REGIONAL UTILIZATION OF REUSABLE PALLETS BY THE GROCERY AND RELATED PRODUCTS INDUSTBY by

Robert Bruce Anderson

Dissertation submitted to the Faculty of the

Virginia Polytechnic Institute and State University

in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

in

Forestry and Forest Products

APPROVED:

MP Raräd Wisdom

—** ‘¤g6m2i‘e„;6;e6*‘—Y"- — JE§xiSF%

Blacksburg, Virginia

to my parents. To my mother whose unfailing confidence and

encouragement has been a keystone throughout my educational career;

all I can say is I am extremely fortunate to have been a part of your

life. To my late father, who this dissertation is dedicated to; all I

can hope is that all the faith he had in me is exemplified in this work.

iii In Memory of my Father

iv TABLE OF CONTENTS

ACKNOWLEDGEMENTS...... ii

TABLE OF CONTENTS...... v

LIST OF TABLES...... vii

LIST OF FIGURES...... viii

CHAPTER I INTRODUCTION...... 1 The Problem...... 3 Objectives...... 6 Organization of Dissertation...... 7

CHAPTER II THE PALLET INDUSTRY...... 8 Industry Development...... 8 Product Standardization...... 10 Grocery Pallet Consumption...... 12

CHAPTER III THE GROCERY INDUSTRY...... 16 Structure of the Industry...... 19 Relationship Between Retail Stores and Distribution Centers....23 Pallet Standardization in Grocery Distribution...... 26 Pallet Use in Grocery Distribution...... 29

CHAPTER IV DISTRIBUTION MODELS...... 37 Review of Literature...... 37 Plant Location Models...... 39 Spatial Equilibrium Models...... N2 Econometric Models...... M7 Simulation Models...... 5M Applicability of Existing Models...... 56

CHAPTER V GROCERY DISTRIBUTION MODEL...... 65 Spatial Aspects of Grocery Production and Distribution...... 65 Grocery Manufacturers...... 67 Grocery Distribution Centers...... 70 Retail Stores...... 72 Pallet Manufacturers...... 73 Pallet Flows in the Model...... 76 Grocery Distribution Center...... 77 Warehouse...... 79 Pallet Outflow...... 81 - Retail Store...... 81 Grocery Manufacturer...... 83 Pallet Repair...... 83 Pallet Discard...... 86 Pallet Inflow...... Q...... 88 Retail Store...... 88 Grocery Manufacturer...... 91 Pallet Repair...... 91 Pallet Manufacturer...... 92 Demand and Supply Aspects of the Model...... 9U Final Product Market Level...... 95 Materials Handling Elements of Final Product Supply...... 99 Grocery Manufacturers...... 101 Grocery Distribution Centers...... 102 Demand for Materials Handling Services...... 10ü Pallet Market Level...... 108 Stock Adjustment Model For Grocery Distribution...... 113 The Complete Model...... 121 Grocery and Related Products...... 12ü Pallets...... 12ü

CHAPTER VI PALLET CONSUMPTION BY THE GROCERY INDUSTRY...... 125 Data Base...... 125 Estimation of Pallet Consumption...... 1üO Conversion Factors...... lüü Regional Pallet Consumption...... 1ü9

CHAPTER VII EFFECT OF GROCERY PALLET USE ON HARDWOOD RESOURCE....156 Wood Raw Material - New Pallets...... 157 Wood Raw Material - Pallet Repair...... 162 Industry Trends in Wood Use...... 168 Continued Pallet Use...... 168 Continued Preference for Hardwood Species...... 169 Increasing Standardization of Pallets...... 170 Captive Pallet Operations...... 173

CHAPTER VIII SUMMARY AND CONCLUSIONS...... 175 Summary...... 175 Conclusions...... 183 Future Research...... 186

LITERATURE REFERENCES...... 189

APPENDIX A...... 197

APPENDIX B...... 198

APPENDIX C...... 200 Web VITA...... 205

ABSTRACT......

-Vi- LIST OF TABLES 1*2% Ess

1 Percent of tons of product movement palletizable, by shipper group and class, for all geographic regions...... 13

2 Regions defined in the 1987 Marketing Guidebook...... 128

3 Food Store Sales in 1986, by Region and Market Area...... 130

N Market Area: Charleston/Roanoke...... 133

5 Retail Sales (X) Attributed to Distribution Centers, 1986...... 135

6 Retail Sales ($000) Attributed to Distribution Centers, 1986...137

7 Pallet Production and Grocery Store Sales...... 1ü2

8 Regional Sales, as a Percent of Total Sales in 1986...... 151

9 Estimated Regional Pallet Consumption in 1986...... 15N

10 Estimated Wood Raw Material Consumption in Grocery Pallet Production, 1986...... 160

11 Estimated Wood Raw Material Consumption in Grocery Pallet Repair, 1986...... 167

vii LIST OF FIGURES

Figure Qggg

1 Spatial Aspects of Grocery Distribution...... '...... 68

2 Flow of Pallets in a Grocery Distribution System...... 78

3 Pallet Outflow from a Grocery Distribution Center...... 82

U Pallet Inflows to a Grocery Distribution Center...... 89

5 Demand and Supply Linkage Between Market Levels...... 96

6 Comparison of Bureau of the Census and Marketing Guidebook Regions...... 127

viii CHAPTER I

INTRODUCTION

The U.S. pallet industry depends greatly on the future availability of timber resources, since over 90 percent of its products are made from wood. Since 1960, annual pallet production has quadrupled, and the industry's use of hardwood lumber has increased from

1N percent to more than 50 percent of total hardwood lumber production.

By 1976, 3.2 billion board feet of hardwood lumber were consumed by the pallet industry out of a total production of 6.5 billion board feet

(USDA Forest Service 1982). The increased demand for pallets increases the volume of wood consumed by the industry. Between 1960 and 1979, wood use by the pallet industry has increased an average of 7 percent per year. Present hardwood consumption for pallet products includes raw material inputs such as cants, short bolts, and tree—length logs as well as hardwood lumber. Thus, the total hardwood input to pallet production is greater than just 50 percent of hardwood lumber production.

Growth in pallet usage is derived from two sources: increases in industrial production and increases in movement of domestic products on pallets. Increases in movement on pallets results from manufacturers and distributors converting to pallet handling systems from other material handling systems. Increases in industrial production results in an increase in pallets needed for systems already using pallets.

1 2

From 1950 to 1960, average growth in pallet production per year

was Ü2.1 million units, of which 3.8 million was due to growth in

industrial production and 38.3 million was due to palletization (Wallin

and Luppold 1983). For the period 1961 to 1983, average growth in

annual production attributed to industrial growth was 7.6 million while

that attributed to palletization was 152 million out of a total of 160

million.

The grocery and related products industry is a potential major

growth area for pallet usage over the next decade. This industry is

already the largest market for industrial packaging, which includes

pallets, boxes, and containers, accounting for almost 37 percent of

total packaging in 1980. Its projected growth in manufacturers' output

is expected to exceed 2.8 percent per year between 1980 and 1995 (Walsh

1981).

The grocery and related products industry is unique in two ways.

First, a majority of the products handled by the industry's distribution system are already handled on pallets; and second, these pallets are generally a standardized ü8”xU0" size. In a 1977 survey of pallet manufacturers, the #8xßO pallet constituted 27.5 percent of the total number of pallets produced. McCurdy and Ewers (1985) estimated that in

1982, 20 percent of the pallets manufactured were ü8”xN0" in size, while the next seven most frequently manufactured sizes together accounted for only 28 percent of the total. 3

THE PROBLEM

Future timber demands by the pallet industry are not well

understood. Information required to make decisions based on the

regional levels of these demands is lacking. This has an impact on the

accuracy of timber demand and supply modeling as well as on the ability

of pallet producers to make decisions regarding future availability of

raw material resources for pallet production. There is a need to

determine what levels of pallet demand can be expected within specific

regions and to assess the potential effects of the demand for pallets on

the timber resource within the regions.

' If present pallet use and construction methods continue, the

resulting demand for pallets will have an uneven effect on the

availability and price of the hardwood resource in the various regions

of the country. Some regions of the country have large quantities of

underutilized hardwood resources, and these regions, theoretically, can

meet the needs of increased demand without significant raw material cost

increases. Other regions are utilizing a much larger portion of the

existing hardwood resource and a large increase in demand could cause

increases in raw material costs (Anderson 1986, 1987). Because raw

material costs represent a substantial portion of the production cost of

pallets, those regions that can provide lower cost raw material could

have an advantage over resource—deficient regions.

In order to assess the pallet industry's impact on the hardwood

resources, better information is needed on future usage of reusable H pallets by the grocery and related products industry. This industry is

the biggest user of reusable pallets and is expected to continue this

trend as a major user in the future. Knowing current pallet use by this

industry will enable us to make estimates of future use under different

food flow and pallet durability assumptions.

Pallets have a relatively low value—to-weight ratio, so their

initial use usually occurs within the same region in which they are produced. In a recent survey of pallet manufacturers, McCurdy and Ewers

(1986) found that firms sold most of their pallets within a 100 mile

radius of the plant. The median distance was only 50 miles, with 61 percent of the firms selling pallets only within their home state. An

analysis of the regional use of reusable pallets, therefore, would be more useful than a national study. Previous studies (Anderson

1986,1987) have shown that sufficient wood resources should be available

in the future at the national level for pallet production, but individual regions may experience demands that exceed their regional resource availability.

Information is needed on how new and used pallets are distributed within and between regions and what effects the interregional movements of grocery and related products have on pallet requirements within regions. The absence of this information will result in market uncertainty, misallocation of resources among regions, reduced quantities traded, and increased equilibrium prices within regions. 5

The beneficiaries of this research include resource policymakers who must anticipate future timber demands by the pallet industry in order to develop long-term forest policy which will facilitate more efficient forest management practices. This research will also benefit pallet producers by providing information that will allow them to better understand the long—term potential and long-term trends in the grocery pallet segment of their markets. The ultimate beneficiary is the consumer, who will benefit from better forest planning and better allocation of forest resources in the overall pallet industry.

Although the pallet industry is the largest single user of hardwood raw material, it is made up of many small, independent firms.

No individual firm within the pallet industry has the resources to do market research which can provide the detailed information required to make informed decisions. The National Wooden Pallet and Container

Association, with which many of the firms have contact, also has a limited budget and staff and cannot provide this type of research. The present study will provide a basis for further assessments of regional pallet usage within other markets. With more complete knowledge of the grocery pallet market and the influence that important factors have on this market, pallet producers will be better able to anticipate future changes in regional market behavior. This information will therefore facilitate better management decisions on the part of pallet producers that could not be made in the absence of such information. 6

OBJECTIVES

The ultimate objective of this research is to provide information that can be used to understand the long-term potential and long-term trends in the grocery pallet market as they relate to future regional timber demands by the pallet industry. Specific objectives within this overall objective include:

1. Provide information on current use of grocery pallets in the grocery distribution industry through the identification and quantification of grocery pallet use within the retailing and wholesaling sectors of the grocery and related products industry.

2. Provide a theoretical framework for future analysis of the regional demand for grocery pallets resulting from the use of grocery pallets in satisfying grocery distribution between regions and between market areas within the same region, and determine the relationship between grocery pallet use and regional grocery pallet demand under specific food flow and pallet durability assumptions.

3. Provide information on the demand for regional timber resources resulting from grocery pallet production within specified regions. 7

ORGANIZATION OF DISSERTATION

The pallet ~arket is described in Chapter II. The grocery and related products market is described in Chapter III, addressing the requirements of the first objective. The information presented in

Chapter II and Chapter III, along with a review of literature presented in Chapter IV, is incorporated into the model development presented in

Chapter V. The estimated demand for pallets by the grocery and related products industry demand, the second objective, will be presented in

Chapter VI. The relationship between pallet demand and the regional timber resource, the third objective, is presented in Chapter VII.

Chapter VIII is a presentation of the summary, and conclusions of the study. ° CHAPTER II

THE PALLET INDUSTRY

This chapter describes the major characteristics of the pallet industry, beginning with a summary of the development of the pallet and

its use in modern materials handling systems. Emphasis is placed on the market for hardwood pallets, since hardwood pallets are preferred for

the distribution of grocery and related products. The degree of standardization within this market and possibilities for substitution will be discussed.

INDUSTRY DEVELOPMENT

A pallet has been described as ”...a low, sturdy platform on which materials in process of manufacture, or finished goods, may be stacked in order to expedite their handling, movement, and storage with the use of mechanical fork-lifts, and/or hand trucks" (Panshin, et al.

1962).

The wood pallet industry is composed of about 2500 firms producing a wide variety of wood pallets, skids, bases, containers, and dunnage items. The 1977 Census of Manufacturers shows less than 27 percent of the firms with more than 20 employees (U.S. Department of

Commerce 1981). In a study of the Eastern hardwood pallet industry,

Sendak reported nearly 50 percent of the firms employed fewer than 10 persons (Sendak 1971).

8 9

The wood pallet industry started as a part of the wooden box industry in the early 1920's. The development of materials handling systems using ~echanical lifts required a base or platform on which the goods could be stacked. Military use of pallets prior to and during

World War II provided the motivation for industries to develop palletization programs involving the mechanical handling of goods on pallets (Panshin et al. 1962).

Pallets were initially produced using a table to assemble the individual pieces and a hammer to nail the pieces together. This hand-nailing method of assembly was used by almost all pallet manufacturers through the mid-1950's (Eichler 1976). As Eichler points out, "... in those days the quantities of pallets ordered by customers were smaller and, therefore, change—over time from one size to another was a consideration. Labor rates were much lower and fringe benefits were nonexistent; therefore, production output per man was not of too much concern."

As production needs increased, the technology of assembling pallets improved. Pneumatic gun nailers and stapling machines replaced the hammer. Using these devices and an assembly table enabled production workers to more than triple production rates that could be expected from the hand nailing method of assembly.

Further increases in the demand for pallets resulted in the development of pallet-nailing machines or pallet assembly lines. These machines provide semi- or fully-automatic assembly of pallets at twice 10

the rates obtained with the pneumatic gun nailers and stapling machines. Built—in assembly tables on these machines automatically position the pallet parts for nailing or stapling. By combining the pallet—nailing or stapling machine with conveyors, automatic stringer

and deckboard in-feed hoppers, and automatic pallet stackers, a completely automated assembly line is capable of producing U,000 pallets per day using two production workers.

PRODUCT STANDARDIZATION

The pallet industry is characterized by great diversity in the design, styling, and construction of pallets produced. In a 1977 survey, 77 different designs of pallets were reported (NWPCA 1980). In a study of the materials handling environments of 88 warehouses

throughout the United States in 1980, over 100 different designs of pallets were found (Goehring and Wallin 1980).

Pallet life is described by such terms as durable, permanent, warehouse, expendable, shipping, one-way, and reusable; but, in fact, pallet life is a continuum varying from single—use, one—way pallets to long—life pallets that may be used for several years. Between these extremes, the life expected depends on the pallet user's specification.

Pallets are needed to handle materials or products through a series of shipping and/or storage operations. Pallets were initially used for moving and storing goods within a single plant or site. Over 11

the past 3 decades, however, shipping between plants or sites has become a more important function of pallets (Strobel and Wallin 1969).

The Census of Transportation provides data on product movement in the United States by type of carrier, type of product, and geographic region. Nearly 1.5 billion tons of product were shipped inter—city in

1972 (U.S. Department of Commerce 1976). About 20 percent, or 300 million tons, could be shipped by pallet, which is the equivalent of approximately 600 million pallet loads (Wallin 1977). This tonnage represents the potential level of shipments by pallets in 1972. Actual pallet production in 1972 was reported as almost 155 million units

(NWPCA 1983). With an unspecified number of pallets already in the system in addition to those produced in a given year, it is apparent that an individual pallet is used to make more than one trip per year.

It has been estimated that, on the average, a pallet transports six loads per year (Wallin 1977).

Continued pressure to reduce materials handling costs will result in increased palletization in those industries that can, but may not presently, move goods by pallet unit-loads. Mckeever and

Dickerhoof(1980) point out that, ”Rising labor costs, coupled with improved materials-handling systems in warehouses and transportation systems, will directly influence a trend toward increased palletization."

The pallet industry produces a differentiated product for many different buyers. Wallin (1977) points out that a pallet seller may 12

j produce the same product for different buyers, different products for

the same buyer, or different products for different buyers. In each

case, the seller is selling one product--pallets. The buyer

differentiates the product based on the buyer's needs. Thus, the buyer

can substitute pallets from different sellers only if the pallets have

the same design, style, and construction specifications.

GROCERY PALLET CONSUMPTION

The grocery and related products industry offers more

possibilities for substitution of pallets from different sellers because

of the much greater degree of standardization of pallet specifications

within this industry. This means that the same pallet design will be

used to carry an almost endless variety of products. For example, the

typical food distribution center may contain more than 13,000 different

food and related products, all of which can be placed on the standard

N8xü0 pallet.One

problem in estimating grocery pallet consumption is

identifying how much of a given product is actually palletized (Table

1), and more specifically, how much of a given product (identified by

weight, volume, or number of items) is carried on an individual pallet.

In a study which detailed information on 2706 shipments by M22

manufacturers to 10 major food distribution centers, Strobel and Wallin

(1969) classified products on the basis of physical handling 13

Table 1.-- Percent of tons of product movement palletizable, by shipper group and class, for all geographie regions.

Code Shipper Group and Class Description Percent No.

012 Meat products 100.0 013 Dairy products 100.0 021 Canned fruits and vegetables 100.0 022 Canned speeialities, seafood, frozen food 100.0 023 Grain mill products, eane and beet sugar 8.6 02ü Miscellaneous and kindred food products 100.0 031 Candy and confectionery 100.0 032 Alcoholic beverages 87.0 033 Canned & bottled soft drinks and flavorings 100.0 03N Tobacco products 88.0 06- Paper and allied products 37.9 071 Inorganic chemicals, gases, dyes, and pigments 8.6 072 Miscellaneous industrial ehemicals 8.ü 073 Plastics, synthetic resins, rubber, fibers N9.2 082 Soap, detergents, perfume, eosmetics, etc. 100.0 083 Pain and allied products 13.0 08N Wood, agricultural, miscellaneous ehemicals 8.6 10U Miscellaneous plastics products 50.0 13- Stone, clay, glass (less containers) products 51.6 132 Glass containers and other products 100.0 171 Metal cans 100.0 172 Bolts, nuts, screws, rivets, washers 100.0 21- Electrical products and supplies 21.3

Total, all products 19.9

Source: Wallin, W. B. 1977. Characteristies of the U.S. Pallet Industry. Unpubl. Rep., Forestry Sciences Lab., Princeton, W. Va. lü characteristics. Their classification resulted in the following 5 product groups:

Group I--paper products Bags, meat trays, toilet tissue,facial tissue, towels, napkins. Group II--low-density items Baking soda, cereals, charcoal, crackers, cookies, dried fruit, pet food in bags, toys, games. Group III--canned goods Evaporated milk, fish, fruits juices, pet food, pork & beans, soups, spaghetti, vegetables. Group IV--products in glass Baby food, catsup, jams, jellies, _ spreads, salad dressings, shortenings, oils, syrups. Group V-—heavy package goods Baking mixes, flour, powdered milk, rice, salt, sugar, soaps and detergents.

One factor that influences grocery pallet consumption is change in the consumer's utility functions. Changing economic status (e.g., multiple wage-earner households) and changes in family eating habits have increased the demand for more pre-processed and convenience foods

(Conner, et al. 1985). For example, the demand for instant potatoes and hamburger helper has increased faster than the demand for raw potatoes and meat products which require more processing in the home prior to consumption. Because of the additional processing, which is accomplished at the food manufacturing facility as opposed to in the home, the grocery product is more frequently packaged in a manner which lends itself to palletized handling of the product.

Changes in the processing levels of meat products at the primary processor is another example of a grocery product which has been altered to lend itself more to palletized handling. Ten years ago, much of the 15 meat handled at grocery distribution centers was in the form of hanging carcasses, which were processed into individual cuts of meat at the retail stores. Presently, the carcasses are broken down at packing houses, boxed by grouping major cuts of meat, and shipped on pallets to the distribution centers. The palletized boxes are distributed to retail stores where the individual cuts of meat are finally packaged for sale to consumers. CHAPTER III

THE GROCERY INDUSTRY

This chapter outlines the major characteristics of the grocery and related products industry. These characteristics include the structure of the industry, standardization of pallets within the industry, and current industry practices with regard to pallet use. Particular emphasis is given to the distribution system between the wholesale and the retail segments of the industry. Pallets are used within the distribution system from grocery manufacturers to retail stores. Demand for grocery pallets is derived solely from the need for movement and storage of grocery and related products within the industry. The above characteristics of the industry must be considered as possible influences on the long—term potential for grocery pallet demand.

In order to provide resource policymakers and pallet producers with information on the current use of grocery pallets in grocery distribution, a survey of the distribution system was conducted.

Information was obtained through personal interviews with distribution center managers, both on-site and by telephone contact, and through on-site inspection of distribution center warehouse operations. The format of questions asked during each interview was informal rather than in the form of a formal, or structured, questionnaire.

Two reasons existed for using an informal format for questioning distribution center managers as opposed to using a structured

16 17

questionnaire. First, because the US Forest Service provided the funds

for the on-site visits to distribution centers, government regulations

restricted the conduct of interviews based on a written questionnaire.

Prior approval for written questionnaires was required. Since the

approval process for questionnaires normally exceeded 12 months with no

guarantee that all pertinent questions would appear on the approved

questionnaire, an informal format was considered to be the best choice.

Second, from earlier contacts with industry members, it was clear that a

mailed questionnaire to distribution management would seldom reach the

individual with the most detailed knowledge of pallet use. In fact, in

all nine on-site personal interviews at least two individuals, and in

one case three individuals, were interviewed before the informal series of questions could be completely answered.

Although the questions asked in each interview were informal, the

same areas of interest were covered in each interview. Following each interview, notes on the responses of distribution center personnel were collected and follow-up telephone calls were made to clarify any information that was left out during the initial interview.

In presenting the results of these interviews in this and subsequent chapters, I have attempted to show the consensus response from all the distribution center personnel that were interviewed. This information is indicated as coming from industry sources and should not be considered as an average of all the responses received. Rather, the 18

information presented reflects the respondents best estimates of

industry operation levels at the current time.

The following chain and independent grocers' distribution centers

were visited:

Food Lion Stores, Salisbury, N.C.

Giant Food, Inc., Landover, Md.

The Kroger Co., Salem, Va.

The Kroger Co., Charleston, W. Va.

The Kroger Co., Cincinnati, Ohio

Richfood, Inc., Mechanicsville, Va.

Safeway Stores, Inc., Landover, Md.

Virginia Foods of Bluefield, Bluefield, Va.

Winn-Dixie Stores, Inc., Charlotte, N.C.

The following companies were contacted by telephone:

Acme Markets, Inc., Philadelphia, Pa.

Albertson's, Inc., Boise, Idaho

American Stores Co., Salt Lake City, Utah

Atlantic & Pacific Tea Co., Montvale, N.J.

Giant Eagle, Inc., Pittsburgh, Pa.

Lucky Stores, Inc., Dublin, Calif.

Publix Super Markets, Inc., Lakeland, Fla.

Supermarkets General Corp., Woodbridge, N.J.

These contacts provided the information used in the analysis of the use of pallets in the food distribution system. Distribution 19

centers visited include a representative sample of the top 20 food

chains, as ranked by sales, several independent food retailers, a

cooperative, and an independent food wholesaler. As a group, the food

retailing chains and independents included in this survey combine to

account for over 50 percent of all retail food sales in the United

States.

STRUCTURE OF THE INDUSTRY

The present study is directed at an analysis of the distribution

segment rather than the manufacturing segment of the grocery and related

products industry. Handy and Padberg (1971) point out that distribution

is typically handled by the manufacturer of products and that no models

existed prior to their analysis concerning behavior within an industry

dominated by a distributor. They further point out that large

distributors occurred only infrequently outside the food industries

until recently and that the economic characteristics of the large

distributor have not been extensively studied.

Handy and Padberg's model of competitive behavior in the food

industries provides a basic description of the structure of the

industry. It includes a model of bilateral interaction (large manufacturer vs. a large distributor) with functional specialization among manufacturing and retailing sectors. Their model is specified schematically rather than quantitatively. The analysis includes the identification of manufacturing, distribution, and retailing sectors 20

within the food industry. Behavior patterns are found to result primarily from the tendency toward specialization by the various sectors.

The structural elements of Handy and Padberg's model include a food manufacturing oligopoly core; a large fringe of small- and medium- size food manufacturing firms; a food distribution oligopoly core (large food chains); and, a large fringe of small- and medium-size food retailing firms. These structural elements are defined as follows:

Core manufacturers-these food manufacturers are large firms, usually diversified into many products. A major part of their competitive strategy revolves around improving brand meaning and impact.

Core distributors-the primary competitive advantage of core distributors is their preretailing operations. All core distributors have warehouses, manufacturing plants, quality control labs, and computer-controlled logistic systems. The preretailing advantages pertain to matters of cost and efficiency, not product quality.

Fringe manufacturers-these many small- and medium-size food processors have little or no marketing capability. Private-label programs enable them to specialize in the physical functions of — food processing their primary competitive advantage.

Fringe distributors- these do not have preretailing capabilities of the big chains. Their advantages involve greater merchandising flexibility. Smaller retailers use this flexibility to adapt their stores to the particular needs of communities they serve. The wider variety of more progressive products from core manufacturers fulfills their product and service needs. In competing with the standard offerings of large chains, fringe distributors have become more innovative in store design as well as merchandise variety. In this way, their competitive advantage is compatible with the competitive emphasis of core manufacturers (Handy and Padberg 1971). 21

Core distributors tend to emphasize private label programs; hence,

they are best served by the fringe manufacturers. This combination is a

separate channel organized to emphasize physical efficiency and price

competition. The drive for physical efficiency directs innovative

activity within this channel toward process development as opposed to

product development. Products within this channel tend to be

differentiated on a price basis.

Core distributors have a comparative advantage in pre—retail

operations which are subject to significant economies of scale. Most

warehouse economies are realized by operations of $100 million or more

annual retail sales. However, to obtain economies of manufacturing and

private-label operations, annual retail sales of over one—half billion

dollars may be necessary.

Fringe distributors tend to work most directly with core

manufacturers. This combination of large, diversified-product

manufacturers and smaller, more specialized distributors constitutes a

channel that emphasizes innovation and progress in defining the

character of the product and services. It is on this basis that products within this channel are differentiated.

Fringe distributors are able to overcome much of the core distributors' pre—retail cost advantage by more effective performance at the retail level. Progressive fringe distributors rely on unique store decor, highly motivated personnel, and innovative merchandising programs. 22

While distribution oligopolies have achieved their most advanced development in food industries, they now extend far beyond this sector, particularly into the general merchandising discounting field.

Distribution oligopolies are probably a structural characteristic of the more mature industry sectors (Handy and Padberg 1971).

A firm may be said to possess market power if a price, production, marketing, or purchasing decision it might practically make can directly and materially affect the incomes of other firms or persons or can appreciably change the average price, total quantity, or marketing or purchasing practices in a market in which it participates (Brandow

1969). Under oligopoly, rivalry is personal, firms have character, power is part of the industry environment; under pure competition, everything focuses impersonally on price, and producers may not even regard each other as rivals.

Because food manufacturing and distribution is considered to be dominated by a few large firms, the long-term potential for pallet demand depends more on the continued growth in retail sales than on changes in industry operation techniques. In a mature industry with substantial capital investment in materials handling equipment dependent on the use of pallets, substitution of alternative materials handling techniques that would completely replace the use of pallets would involve large expenditures of new capital. 23

RELATIONSHIP BETWEEN RETAIL STORES AND DISTRIBUTION CENTERS

Any retail store selling a line of dry grocery, canned goods, or

non·food items plus some perishable items is defined as a grocery

store. The number of grocery stores has varied over the years.

According to statistics published by Progressive Grocer Information

Sales, in 1985 there were approximately 15ü,000 grocery stores in the

United States, broken down as follows:

All grocery stores 15ü,000

Supermarkets (over $2 million) 30,505

Independent Supermarkets 13,285

Chain Supermarkets 17,220

Convenience Stores #5,üOO

Other (small) stores (under $2 million) 78,095

Over the last NO years, there has been a decrease in the number of retail grocery stores in the United States from over 3ü0,000 to less than 155,000. Most of this decrease was in stores operated by sing1e·unit firms. The number of stores operated by multi—unit firms has increased substantially, particularly for firma operating more than

11 units, which is defined as a chain. Also, the growth in convenience stores has been dramatic. Convenience store sales in 1980 were $2U.5 billion and $N7.5 billion in 198N, nearly doubling.

The primary function of a grocery distribution center is to act as a central point from which a retail store can receive products to replenish the products sold. It would be prohibitively expensive for zu

individual retail stores to purchase products directly from a

manufacturer in the small quantities which a retail store deals with on

a daily or weekly basis, primarily because of the shipping costs

involved. Also, the size of most retail stores prohibits the

stockpiling of large quantities of goods to meet consumer needs for an

extended period of time.

The distribution center therefore provides an intermediate buffer

between the manufacturer and the consumer, allowing purchases of large

quantities of goods at substantial volume discounts and the storage of

those goods until they are needed by the retail store. Also, the

distribution center allows the development of a more efficient

distribution system which involves the use of pallets for moving the

products to the retail store than would be possible if all retail stores

had to receive products directly from manufacturers. Instead of many

trucks making deliveries of small quantities of products, one truck can

make deliveries of a multitude of products to the same retail store.

Four categories of grocery distribution centers may be

identified: chain—store distribution centers; voluntary-group

distribution centers; cooperative distribution centers; and

non—sponsoring wholesale distribution centers. The distribution centers

in each category operate independently. That is, there is no trading of pallets between centers in different categories. The chain-store distribution centers generally serve a national or regional chain of retail stores. The voluntary-group distribution centers act as 25

wholesale sponsors for a voluntary merchandising group of independent

retail stores who operate under a common company name. The cooperative

distribution centers serve generally independent retail stores who are

stock holding members of a cooperative wholesale buying group. The

non-sponsoring wholesale distribution centers serve retail grocers who

are unaffiliated with any of the other distribution centers.

One feature which distinguishes the above categories of

distribution centers is in the corporate relationship between the

centers and the retail stores they serve. The chain—store distribution

center is generally a separate profit center within a larger

corporation. The other three types of distribution centers are

corporate entities in themselves, although in the case of the

cooperative distribution center, the retail stores do own stock in the corporation. Another feature is the size differences between distribution centers in each category. This relates to the market share held by each category. Generally, the chain-store distribution centers are the largest in terms of physical volume of products moved through them on an annual basis, while the non-sponsoring wholesale distribution centers are the smallest. The other two catagories fall in between these two extremes. The larger distribution centers are more likely to depend on automated handling equipment within the center to handle to the volume of product moving through the center, while the smallest centers are more likely to use a greater amount of manual labor to move products through the center. 26

Production and distribution economies of scale are substantial in

the food industry. As noted earlier in the section on the structure of

the industry, Handy and Padberg (1971) estimate that to obtain economies

of manufacturing, distribution, and retailing, annual retail sales of

over one—half billion dollars may be necessary. With regard to just

grocery distribution, they estimate that most economies of scale are

achieved when annual retail sales handled by distribution centers equals

or exceeds $100 million. This is true whether the grocery distribution

center is a part of a larger corporation or an individual wholesaler.

The optimal size for a distribution center is determined by the size of

the market served, which relates to the number of retail stores served

and their level of retail sales, and the distances involved between the

retail stores and the distribution center. Current structural changes

in the grocery distribution industry, such as increasing access to

scanning data, introduction of high-rise storage facilities, and the

development of better pallet handling equipment, will further increase

the economies of scale in grocery distribution. As grocery distribution

centers achieve these economies of scale, their pallet procurement power

should increase more rapidly relative to grocery manufacturing.

PALLET STANDARDIZATION IN GROCERY DISTRIBUTION

Grocery distribution centers handle a wide variety of products.

These products include dry grocery products, canned goods, produce,

frozen food, meats, and dairy products, as well as many types of 27

non-food items. Because of federal and state health and sanitation

regulations, all products stored in a distribution center warehouse must

be stored on some type of platform or, at least, not directly on the

floor.

Based on the survey of distribution centers, it is concluded that

nearly all merchandise in the grocery distribution warehouses are on

pallets. Most frequently, pallets are used to store merchandise in

pallet racks. Nothing is stored directly on the floor, although floor

stacks of goods on pallets are not uncommon. When floor stacks of goods on pallets are found, the maximum height of the stacks seldom exceeds U pallets in height. Floor stacking is used exclusively in few warehouses. It is used primarily for holding inventory of large volume,

low unit-weight items, although a few warehouses do pick orders for shipping from floor stacked pallets.

Materials are handled on pallets in warehouses with few exceptions. When goods arrive at the warehouse on slipsheets or deadpiled on the floor of a truck or railcar, they are placed on pallets at the receiving dock before they are moved to storage in the warehouse. Goods arriving on nonstandard pallets are restacked on warehouse pallets prior to entry into the warehouse. Gravity-fed hoppers may be found in smaller warehouses where low-volume items (less than case lots) are available to pickers, one unit at a time.

The standard grocery pallet size used in warehouses is N8 by Mo inches. Although different pallet sizes may be found, the use of 28

pallets other than the standard N8 by NO pallet is limited to special

applications. These include dairy operations, health and beauty aids

sections, slow moving items sections, automated meat lines, and

automated frozen food sections. Typically, these special pallet sizes

are found in captive pallet systems, that is, in systems where the

pallet never leaves the warehouse or where control of the pallet is

maintained by the distribution center.

No distribution center could function without the aid of a

computer. Items are received, stored, picked, and shipped as a result

of a computer generated order. Use of computerized coded bins to store

goods within the warehouse means that the space in the warehouse is

divided up into a number of cubes, each one of which is identified by

the computer, so that any individual cubic space in the warehouse is

identified by an appropriate code and any good stored in that space can

also be identified by an appropriate code.

Maintaining the quality of pallets in the system is a major problem for distribution center managers when pallets are exchanged with

suppliers and are used in the distribution to retail stores. In a direct exchange of pallets between distribution centers and suppliers of grocery products, the pallet received under product and the empty pallet returned to the supplier must be equal in quality. This requirement is not always satisfied. No distribution center reports receiving the same quality pallet in the direct exchange, although managers maintain that the distribution centers return only good or satisfactory pallets. 29

Pallet standards do not appear to be substantially different in any of

the distribution centers visited. Causes of damage to pallets have been

explored elsewhere in the literature, but this still remains as a source

of concern throughout the food distribution system.

PALLET USE IN GROCERY DISTRIBUTION

The life of a pallet in the food distribution system is limited.

At some point, pallets must be replaced so that a minimum level of

pallet quantity is maintained. Pallets may enter a distribution center

in a number of ways:

1. Direct purchase from a pallet manufacturer.

2. Direct purchase from a pallet distributor.

3. Indirect purchase from a pallet manufacturer.

N. Indirect purchase from a pallet distributor.

5. Exchange with grocery vendor.

6. Exchange with other distribution centers.

7. Exchange with retail stores served by the distribution center.

Considerable variation exists among distribution centers methods for maintaining a minimum level of pallet quantity in the system.

However, price and availability appear to be the governing factors in the decision regarding how the level of pallet quantity is to be maintained.

The quantity of pallets purchased directly from either a pallet manufacturer or pallet distributor varies considerably from one 30

distribution center to another. Few distribution centers purchase all of their pallet needs directly. In these cases, local pallet manufacturers or pallet distributors are contacted and asked to submit bids for the delivery of pallets meeting the distribution center's specifications. Some distribution centers purchase only new pallets while others purchase only reconditioned or used pallets. Price and availability of used pallets are critical in the case of used pallet purchases.

Most distribution centers obtain pallets through indirect purchase from pallet manufacturers or pallet distributors. That is, they negotiate with the manufacturers of grocery products, hereafter noted as vendors, to deliver their grocery products on pallets which meet the specifications required by the distribution center. These may be new pallets which the vendor purchases from a pallet manufacturer or they may be reconditioned or used pallets purchased from a pallet distributor. The vendor ships the product to the distribution center on pallets and bills the distribution center for the cost of the product as well as the cost of the pallet. These pallets then enter the distribution system and are added to the stock of pallets in the system. This is the predominate way in which new pallets enter the system.

Most vendors who deliver their product to a distribution center on pallets do not have agreements which include the purchase of the pallet as well as the product by the distribution center. In a majority of 31

cases, the vendors shipping on pallets have a direct exchange agreement

with the distribution center. The vendors will deliver goods on pallets

to the receiving dock of the distribution center and pick up the same

number of pallets to take back to the vendor. While the number of

pallets at the distribution center has not changed as a result of this

direct exchange, the composition of pallets has changed. That is, the

distribution center may no longer have exactly the same pallets, in

terms of quality, that it had prior to the exchange. When multiplied by

the total number of similar shipments received, the number of pallets

exchanged can exceed the number of pallets in the distribution center in

a relatively short time period, or in most cases, in less than one month. This means that the quality composition of pallets in the distribution center is very dependent on vendors shipping goods on pallets that meet the standards set by the distribution center, regardless of whether the pallets are to be purchased by the distribution center or are to be exchanged for a like number of pallets.

Some products are delivered to the distribution centers either on slip sheets or on the floor of the truck. The quantities of product received at individual distribution centers in this fashion varies considerably. Some distribution centers receive as much as Mo percent of their products in this way; but, the industry average must be closer to 35 percent. Depending on the arrangement with the individual carrier, the truck driver may be responsible for unloading the truck and placing the product on pallets at the receiving dock. Regardless of the 32

arrangement, all products received on slip sheets or on the floor of the

truck are placed on pallets prior to being placed in storage.

Although pallets and slip sheets require alternate handling

equipment, they may be used together in a distribution system. That is,

products carried on slip sheets may be either placed directly on the

slip sheets, or placed on pallets after being placed on slip sheets.

The latter procedure occurs frequently in the handling of grocery

products where the products are shipped from the manufacturers on slip

sheets and are placed on pallets upon arrival at a distribution center

for subsequent handling in the center. In this case, slip sheets are

complements used with pallets not substitutes for pallets.

The possibilities for replacing pallets with slip sheets in

materials handling of grocery and related products are limited by the

overall handling environment for grocery and related products. Grocery

distribution centers have substantial capital investments in handling

equipment which is designed to accomodate pallets. In order for pallets

to be replaced by slip sheets, distribution centers would be required to make major outlays of capital for new equipment. Thus, the price cross-elasticity for substitution of slip sheets for pallets must be very low. Conversely, the price cross-elasticity for substitution of pallets for slip sheets may not be so low because of the greater flexibility of handling equipment in the area of distribution where slip sheets are presently in use. 33

Exchange of pallets between distribution centers within the same corporation does occur, particularly when one center has an excess number of pallets in relation to its needs and another has a shortage of pallets. This does not occur frequently, primarily because the individual distribution centers associated with a single corporation are usually spread out over the country to provide regionalized service and are not within close enough proximity to each other to be able to afford shipping of pallets between them. A more typical response by a distribution center to an excess of pallets in the system would be to return the excess pallets to the captive supplier, that is, a vendor who ships products on pallets which are kept by the center. In this case, the distribution center essentially sells the pallets back to the captive supplier who may then turn around and ship more product back to the distribution center on those pallets and again bill the distribution center for both the product and the pallet.

Since most distribution centers ship products to retail stores on pallets, another way pallets enter the distribution center is through an exchange of pallets with the retail store. In almost all cases, the retail store submits a request for certain products to the distribution center. This request is translated into a picking order which designates the location of the items in the warehouse, the order in which the items are to be picked, and the quantities of each item that are to be loaded onto each pallet. As the order is picked, the cartons of product are stacked on a pallet in such a manner that the stack has 3ü

— square sides or the pallet is said to be cubed. The top layers of

product stacked on the pallet are tied together in various ways · with

strapping tape, string, or stretch wrap applied by hand or by machine.

Pallets with completed orders on them are taken to the shipping

docks where they are loaded onto trailers for delivery to the retail

lstores. Some distribution centers deliver products to the retail stores

in trailer lots, that is, all of the products on a trailer go to one

store. Other centers will have mixed loads with a trailer dropping off

part of each load at several stores. The latter case occurs most

frequently for centers serving smaller retail stores which do not have

the volume of product moving through the store that the large super

stores have.

The pallets are unloaded at the retail stores, with the products

still on them. Stores load empty pallets, paper bales, bread trays,

milk trays and so on into the trailer which returns to a receiving dock

at the distribution center where the pallets are sorted out and returned

to the system.

Although a majority of distribution centers deliver products to

retail stores on pallets, there are alternative methods such as

dead-piling products on the floor of the trucks and shipping products on metal carts rather than pallets. In both of these cases, the pallets used in the distribution center for storage of products never leave the distribution center unless the center has an exchange agreement with a vendor who ships product to the distribution center on pallets. 35

Backhaul operations are an important part of the distribution

system for grocery and related products. It is more efficient as well

as profitable to have a truck on the road carrying a load rather than

empty, so distribution centers schedule the pickup of product for return

to the center whenever possible. A typical case would be as followsz a

truck delivers a trailer load of palletized product to a retail store,

drops the trailer off, and picks up an empty trailer that was left at

the store on the previous delivery. The trailer is taken to a vendor in

close proximity to the retail store and a palletized load of product is

picked up for delivery to the distribution center. Alternatively, the

trailer may unload at the retail store and proceed to the vendor on the

same day. In these cases, the pallets which are picked up at the retail

store may be traded or exchanged with the vendor when the trailer picks

up the backhaul load. This differs from the earlier case were the pallets were returned directly from the retail store to the distribution

center. The use of backhaul arrangements occurs throughout the country although those distribution centers located in the western part of the

United States appear to receive a larger portion of the products at the distribution centers as a result of backhauls than those in the eastern half of the United States. This is a result of the greater distances involved between distribution centers in the west and their retail stores served.

Keeping track of the pallets in a system requires a substantial amount of paperwork on the part of the distribution center management. 36

The problems are compounded when pallets are exchanged that are not of the same quality or standard required by the distribution center. Those systems which have captive pallets appear to have less problems with maintaining standards than those systems which involve many direct exchanges of pallets.

Pallet repair may be contracted with an outside agency or it may be done inside the distribution center. It appears that the decision regarding whether to repair pallets internally or to contract out is made on the basis of least cost rather than convenience. Pallet repair operations vary in levels of complexity from one—man, hand-nailing operations to_sophisticated pallet un-nailing machines with several employees working multiple shifts replacing up to four deck-boards and one stringer per pallet. CHAPTER IV ‘ DISTRIBUTION MODELS

The initial section of this chapter is a review of the use of mathematical models to analyze market behavior. This section is followed by a literature review of previous economic studies of a variety of markets, including the forest products markets. The techniques used in these studies may have application to modeling of the grocery distribution system. These techniques include plant location models, spatial equilibrium models, econometric models, and simulation models. The concluding section of this chapter will discuss the applicability of existing models to analysis of the grocery distribution system.

REVIEW OF LITERATURE

The use of mathematical programming to analyze market behavior has been explored extensively in a number of studies since Samuelson (1952) pointed out that an objective function can be written by restating the firm's revenue and cost functions in the form of a profit function where profit equals revenue minus cost. Samuelson showed that maximization of the objective function guarantees fulfillment of the conditions of a competitive market. While some of the studies have been purely theoretical, his basic idea also has proven useful in the realm of empirical economics, particularly in the context of agricultural

37 38

planning models which may contain rather detailed supply side

specifications. Fromm (1973) illustrates a number of different

approaches to the use of theory in the specification and estimation of

models and provides a discussion on movement from empirical models to

modifications or improvements of theory.

Duloy and Norton (1975) develop a procedure for representing

competitive and noncompetitive structures in linear programming models.

Arbitrarily close approximations to nonlinear forms - in both the

objective function and constraint set - can be made without much loss of

the computational efficiency of the simplex algorithm. Product

substitution effects in demand can be approximated by a linear program.

The demand structure can be transformed to take account of any shift in

demand which can be represented by a rotation of the demand function.

The objective function in these cases again maximizes profit, which is

expressed as the difference between the revenue obtained from selling

activities and the costs incurred in production activities.

Hazell (1979) provides a method for formulating linear programming models in which one or more factors have upward sloping supply schedules, and the prices of these factors are to be endogenously determined at either their competitive market equilibrium values or at the levels set by a monopsonist. The method for achieving these results utilizes the sum, over the relevant factor markets, of the producers' and consumers' surplus, and is an extension of existing methods for solving price endogenous models of product markets. This procedure 39 extends the Duloy-Norton method to obtain the solution to endogenous factor markets in aggregate agricultural production models. In this case, it is the factor supply function that is given exogenously, and the factor demand function which is implicitly contained in the production decision model. The method is first developed using a continuous increasing cost function, and then the special complexities inherent with the use of step functions are considered.

PLANT LOCATION MODELS

Plant location models have been used to analyze the relative efficiency of various plant sites and to select the least cost locations for production and inter-regional transportation. For example, Fedeler and Heady (1976) have developed ten specifications of a linear programming model to jointly select the least cost locations of grain production and interregional grain transportation in the US. Their results suggest that choice of transportation mode and grain flows are sensitive to transportation cost changes and the distribution of exports among ports but the location of grain production is not. Transportation costs in their model are designed to represent transportation operators' costs, not the market price of transportation services, plus grain elevator loading and unloading costs. Although the flow of grain through an elevator may be stable, the location of those who buy from or sell to the elevator may be sensitive to transportation and export changes . NO

Kloth and Blakely (1971) develop a production—distribution model utilizing separable programming to determine the optimum number, size, and location of processing plants that will minimize assembly, processing, and distribution costs of the U. S. dairy industry.

Significant economies are possible under optimum organizations of the dairy industry, but they note that models which minimize industry costs with a single firm in each market overestimate the potential savings from reorganization of the industry. A substantial portion of such savings might be attributed to the economies associated with the establishment of a single firm in the local markets.

Another algorithm has been suggested that operationalizes the

Stollsteimer (1963) model for plant location problems where a large number of plants may enter the optimum solution (Warrack and Fletcher

1970). Also stated as a product distribution model, given m geographically dispersed market demand (Qj) to be supplied from any one or more of N possible plant locations, their model solves for the number of plant locations, n

Focusing on rail shipments, Ladd and Lifferth (1975) developed a transshipment plant location model which was used to determine the number, size, and location of new subterminals, expansions in storage capacity of existing country elevators, the rail network, and the monthly flows of grain from origins to elevators to destinations to N1

maximize joint net revenue of grain producers within a 6 1/2-county

region. Their method of solution extends the Stollsteimer model.

An activity analysis model was developed to determine the optimum

period of production at a chain of sugarcane processing plants and the

optimal regional transport network flows of cane and raw sugar (Ryland

and Guise 1975). Explicit treatment is given to discrete variations in

input quality which affect revenue at each plant location in each time

period. Optimal solutions to three market configurations open to a — multifacility monopolist spatiotemporal quality competition, spatial

quality competition, and pure competition - are obtained.

A variety of interregional linear programming models have been

used to study the optimal location of the cattle feeding industry.

Byrkett, Miller, and Taiganides (1976) performed an analysis to

determine which factors are most influential in determining feedlot

location and thus need to be included in these models. In addition to

traditional factors, consideration is given to the effects of region

definition and regional land use practices. Their results indicate the

importance of feeders, grain, and land use patterns.

Combination of a single-equation location model and interregional

trade analysis into one model has provided an effective tool to simultaneously determine regionally optimal numbers and sizes of processing plants and optimal interregional trading and pricing. The results of an earlier empirical application of a model on the prospective soybean industry in India were reviewed after four years of #2

actual development (von Oppen and Scott 1976). The model was

constructed in two parts: the plant location was determined with the help of a single equation optimization model and the interregional trade of inputs and products was analyzed by means of a quadratic programming model. Included with the model was a flow chart of a spatial equilibrium model for plant location and interregional trade.

SPATIAL EQUILIBRIUM MODELS

Spatial equilibrium models which analyze price, production, and consumption patterns over time have been constructed for a number of industries. Those models included in this review, by no means an exhaustive list, analyze such industries as the Northeast milk market

(King and Ho 1972), the international paper and paperboard market

(Hassan and Wisdom 1982), the lumber and plywood markets of the United

States (Adams and Haynes 1980), the North American paper industry

(Buongiorno and Gilless 1981 and 1983), the international trade of forest products (Buongiorno and Gilless 1983), the New England dairy industry (Kottke 1970), the U.S. apple industry (Fuchs, Farrish, and

Bohall 197#), the North American pork sector (Martin and Zwart 1975), the world sugar economy (Gemmill 1977), the national coal economy

(Libbin and Boehlji 1977), and the world rapeseed industry (Furtan,

Nagy, and Storey 1979).

King and Ho (1972) used reactive programming developed by Tramel

(1965) to solve a spatial equilibrium problem concerning projected milk **3

prices, consumption and production during the period 1965-1975.

Equilibrium prices and trade flows were calculated based on given demand

functions for each consuming area, supply functions for each producing

area, and transfer costs from each producing area to each consuming

area.

Hassan and Wisdom (1982) developed single market analyses of

international trade for three paper products: newsprint, printing paper,

and paperboard. Demand equations were estimated for all three

commodities within regions. Likewise, regional supply equations based on product and input prices were estimated. Cost of shipping products along major trade routes were estimated, although the same regions were not used in the analysis of each product market.

The Timber Assessment Market Model (TAMM) developed by Adams and

Haynes (1980) analyzed a spatial model of North American softwood lumber, plywood, and stumpage markets. Six product demand regions and nine supply regions (including Canada) were included in the model. The model was designed to provide long-range projections of price, consumption, and production trends. Regional processing response functions, which corresponded to Hassan and Wisdom's supply equations, were developed. Regional stumpage supplies, which depended on price and local forest inventories, were also developed. The assumption was made that regional stumpage demand was limited to the local processing sector. UM

In an economic model of the North American paper industry,

Buongiorno and Gilless (1983) used spatial equilibrium methodology to

calculate production, transport and consumption of raw materials,

intermediate products, and final goods traded in the industry. The

model was designed to predict long-term developments of the industry

under various economic and demographic scenarios. The model focused on

seven supply and five demand regions in the United States. The

forecasts for the paper sector resulting from the model were considered

to be compatible with those of the solid wood sector based on the TAMM

methodology.

An adaptation of the previous model to consider international

trade in wood products was presented by Buongiorno and Gilless (1983)

for the trade of newsprint between major importers and exporters in a

relatively aggregated set of world regions. The model featured the

introduction of potential barriers to trade such as tariffs and quotas

and the inertia of trade adjustments. The model determined equilibrium

imports and exports and corresponding prices by maximizing the surplus

value of trade for all countries simultaneously.

A set of recursive relations incorporating linear and quadratic programming formulations were used to handle the temporal and product-use dimensions in addition to the spatial dimension. An application of such a model to the New England dairy industry traced price, production, and consumption patterns over time (Kottke 1970). N5

The concept of recursiveness in multistages may have general

applicability in the design of similar models for other industries.

Fuchs, Farrish, and Bohall (197N) found that analysis of the U. S.

apple industry and its problems requires simultaneous consideration of

its multiple dimensions of space, time, resources, commodities,

production activities, and marketing levels. They constructed an

empirical quadratic programming model incorporating these dimensions and

demonstrated its use for policy analysis by measuring the impact of

alternative size reductions in regional apple marketings on f.o.b. level

industry and regional net returns. The model could also be used in

determining the ramifications of changes in consumer demand,

transportation costs, and marketing margins on such factors as production, prices, and interregional flows.

A quarterly recursive quadratic programming model of the North

American pork sector was constructed to explain spatial and temporal variations in the sector and to evaluate the repercussions of policy changes (Martin and Zwart 1975). An adjustment in the Takayama—Judge

specification of the quadratic programming model was necessary for the inclusion of storage demand relationships. Storage considerations should be included in spatial analyses when storage is an important factor in the market.

U. S. sugar policy was examined in an international context by means of a spatial equilibrium model of the world sugar economy in which various trade barriers were imposed (Gemmill 1977). The solution gave 46

quantities produced and consumed in each region as well as intersectoral

trade and the domestic price in each region. Five policy experiments of

particular interest to the United States were conducted.

A multiperiod spatial equilibrium model of the national coal

economy was developed to evaluate future interregional shifts in coal

production and the investment requirements and sequencing for

exploitation of a nonrenewable, variable quality resource (Libbin and

Boehlje 1977). The objective function of the model was designed to

minimize the total discounted cost of mining, washing, and transporting

coal, reclaiming strip—mined land, and constructing new mines, subject

to constraints on mining equipment availability, known coal reserves,

current mining standards, and the projected demand for coal. Cost

minimization was justified on the basis of firm size, the competitive

structure of the industry, and the behavior of utilities to obtain

competitive bids. Eighteen demand and twenty-one supply regions were

specified with unique points of origin and destination.

A four-region, three—commodity, spatial equilibrium, quadratic programming model of the world rapeseed industry was constructed to measure the impact of the various policy changes (Furtan, Nagy, and

Storey 1979). Excess demand and supply in primary and intermediate markets was illustrated for equilibrium market conditions in a single commodity, multiregion case. The equilibrium price in each market was determined by the difference in transport and tariff charges. The quantities traded, produced, and consumed in each market related to the [V7

price through the specified demand and supply curves. In the general

case of joint production, prices of the commodities were linked through

a marketing margin equation. The objective function was then to

maximize the net social payoff as calculated in all commodity markets.

If a margin equation was employed, then this equation formed a derived

demand in the primary commodity market. Once the supply of the primary

product was known, then the supply of final products could be determined

in each market. The objective function then allocated the final product

between the regions given the respective demand curves.

ECONOMETRIC MODELS

Econometric models are composed of a number of components which

reflect various aspects of demand, supply, and price determination. In

order to isolate price-quantity relationships by statistical means, variables that cause these relationships to shift must be included in

the analysis. One critical aspect of price determination in many of

these models is the linkage between inventory levels and price

adjustments. Even where sufficient data exists for model building, it may be impossible to identify the particular price needed for market clearing and adjustment.

The market model, which focuses on the price mechanism that serves

to clear the market, is the most basic type of econometric model.

Process models, on the other hand, deal with supply and demand within an industry rather than across a market. That is, they focus on the M8

transformation of commodity inputs into finished products. While market

models balance supply and demand to produce an equilibrium price, prices

in a process model are normally a function of production and material

costs. Thus, process models concentrate on the industrial production

process, requirements for raw materials, and labor and plant capacity

(Labys and Pollak 198ü).

Simultaneous equation models are one of the most frequently

developed types of econometric models. For example, the major cyclical

characteristics of the coffee economy have been explained by this type

of model, which considers the lagged response of supply to price

(Edwards and Parikh 1976). Coffee prices and production are

characterized by strong long-term cycles, around which are found smaller

short-run fluctuations. These may be explained to a large extent by a

simultaneous equations model, taking account of the lags in the

responses of demand and supply to prices. Such a model was established

to simulate the impact of alternative stabilization policies, the objectives of which were to improve the average level of earnings, to break the long-term cycle, and to reduce the amplitude of short—term fluctuations.

Mckillop (1967) developed an econometric model consisting of a system of linear supply and demand relationships for a variety of wood products and primary products in the United States. The major aims of this study were to estimate the structural parameters of the relationships and to provide point estimates of the demand and supply '49

elasticities. Coefficients of wood product supply and demand

relationships were estimated using a 2-stage least squares procedure.

Additionally, forecasts of future consumption and price levels were made

for the period ending in 1975.

In contrast to the previous simultaneous equation model of

McKillop, which estimated supply and demand functions for a broad range

of forest products, Adams and Blackwell (1973) developed a model of the

forest products industry at the various stages of production which

appears to be more recursive in nature. They noted that the model

combined features of the process model with elements of a market model.

This model was used to forecast lumber demand, supply, and prices to

1975 and to examine various policies for limiting future price

increases.

Manning (1975) developed an econometric model of the Canadian

softwood lumber industry that explicitly investigated the relationships

among Canadian softwood forest industry sectors, prices, and substitute

goods. While similar to the two previous models, the model was

different in that it included a major export market, the United States,

as a sector in the model. Canadian requirements for softwood lumber were hypothesized as a residual of United States' demand for Canadian

softwood lumber, where that demand was determined by the price and level of construction activity in the United States.

Adams (197ü) focused on the response of prices and output to alternative National Forest timber supply policies in an econometric 50

model of the Douglas-fir Region forest products markets. Using a set of

simultaneous linear equations, the structural relationships between the

major sectors and subsectors of the regional forest economy were

identified. The major sectors included in the model were the stumpage

sector, log sector, and secondary products sector. The latter sector

was similar to that developed by McKillop for the lumber and plywood

subsectors. The third component of the secondary products sector, pulp

products, was treated as exogenous.

The relationships between changes in food sector input costs and

retail food prices have been examined using an analysis based on a

twenty equation econometric model of the food•price determination

process, specified following Popkin's "stage of processing" approach

(Lamm and Westcott 1981). Their results indicate that increases in

factor prices pass quickly to consumers, within two quarters for most

foods.

A logit model was used to estimate the elasticities and cross

elasticities for freight transport services. The model was applied to a

sample of cherry and apple shipments. The performance of the model in

explaining choice of transportation method was found to be highly

satisfactory (Miklius, Casavant, and Garrod 1976). Inventory

considerations as well as the decision where to buy were closely

interrelated with the choice of the transport mode. The logit model has been modified to handle choices among unranked alternatives. The decisions to purchase from different production areas, therefore, could 51

be analyzed concurrently with the choice of transport mode. The authors

limited their analysis to the choices of transport mode.

Another econometric model by Arzac and Wilkinson (1979) was

designed to provide quarterly forecasts for such variables as livestock

and grain production and prices, the retail-producer price spreads for

meat products, and consumer demand for meat. The forecasts were

conditional upon assumed values of such exogenous variables as

disposable personal income, government policy with respect to the

livestock and feed grain markets, and certain other developments in the

economy.

Three independent methodological approaches and data sets were

used to estimate the consumer loss due to monopoly in the US food

manufacturing industries for 1975. They include estimatess- built up

from previously estimated components of consumer loss; - derived from a

regression analysis of the relationship of market structure to industry price-cost margins; and, - derived from regression analysis of the market structure determinants of national brand-private label price

differences. All three estimates converge to the $12 to lü billion range. Virtually all of the consumer loss is attributed to income

transfers; 31 to 61 is due to allocative inefficiency (Parker and Conner 1979) •

Heien (1980) presented a dynamic model of farm and retail prices and quantities. The system derived its dynamics from the assumption that supply and demand were not in balance and that this imbalance was 52 the determining factor in causing price changes in auction-type markets. Central to this theory was the notion that changes in retail food prices were caused by changes in prices at lower levels in the marketing chain. These cost changes were transmitted via markup type pricing rules which were shown to be consistent with firm optimization behavior under the assumption of constant returns to scale and time-fixity of production coefficients.

A dynamic model of the consumer demand for durable goods developed by Houthakker and Taylor (1966 and 1970) uses the concept that current purchases of a consumer durable depend on current income and prices as well as the level of inventory on hand for a particular consumer durable good. In this model, noted as a stock adjustment model, current purchases are viewed as an attempt to maintain inventories at some desired or equilibrium level and are influenced by past behavior. The effect of past behavior is assumed to be reflected in the current quantities of certain ”state variables”, which are changed by current decisions but these decisions depend on all past purchases with more emphasis on recent purchases and corresponding less emphasis on purchases made long in the past.

Other studies by Huang (196ü), Duncan (1980), Berkovec (1985), and

Chow (1957 and 1960), apply the stock adjustment model to the demand for durable goods in the automobile industry. In these studies, the desired stock of automobiles is considered to be a function of the relative price of automobiles and consumer income. Purchases of automobiles are 53 .

affected by differences between desired and existing stocks of

automobiles.

Pearce and Wisley (1983) use the stock adjustment model to develop

a model of retail inventories that measures changes in retail inventory

investment. Inventory investment is shown to be a function of

differences between the desired stock of inventories and the actual

stock held at the beginning of some time period as well as unexpected

sales in the period. The desired stock of inventories is hypothesized

to depend on expected sales and on the expected real rate of interest.

Their findings indicate that retailers make adjustments quite rapidly

when differences exist between actual and desired inventory stocks.

Kmenta (1971) and Gordon (1978) present models of stock adjustment

for intermediate goods, which use the concept that firms will maintain

some functional relationship between the quantity of the durable good

and the firms' expected sales. Gordon's accelerator hypothesis of net

investment uses the concept that firms will maintain a fixed relation between their stock of capital and expected sales. Similarly, Kmenta

suggests that the volume of stock of a commodity that a firm seeks to maintain can be expressed as a function of sales. In both cases, the desired level of stock at the end of period t, or Q*t, and the sales during period t, or Xt, are related as follows:

Q f { } t = Xt

The adjustment made by the firm to the desired level of stock in any one period may be expressed as follows: 5M

Qt Qt_1 = 8 ( Q Qt_1 ) - t - where Qt equals the actual level of stock at the end of period t. The

multiplier, g*, represents the percentage adjustment made between the

actual and desired levels of stock in any one period.

Griliches (1960) uses the stock adjustment model in developing a

demand model for a durable input, farm tractors, to the production of

agricultural products. Two features distinguish this model from other

models which examine the demand for a consumer durable good. First, the

model does not include a "scale" variable like consumer income.

Griliches notes that in the conventional theory of the firm, the firm

has no "budget restraint". The firm's only constraint is its production

function. Second, in estimating the annual purchase of tractors,

replacement demand is assumed to be proportional to existing stock.

Thus, annual purchases equal the sum of changes in actual stock,

resulting from differences between actual and desired levels of stock,

and replacement demand.

SIMULATION MODELS

Simulation provides a fourth approach to analyzing the use of

pallets in the food distribution system. In general, simulation is an

approach which uses a model of a situation or a system and manipulates

it with the help of a computer in order to imitate the system's behavior over time for the purpose of evaluating alternative operating decision rules. A major advantage of using simulation techniques is the 55

flexibility that is possible in the formulation of assumptions about the

system. Relationships in the system do not need to take the linear or

other forms required for analytical solutions.

A simulation model for hog production has been developed as part

of an information system which may assist hog farm managers both in

choosing between competing management strategies and also in

implementing any chosen plan (Blackie and Dent 1976). The process

involved in selecting the most suitable management strategy is

illustrated using data from several hog units. The incorporation of the

model into an information system ensures that the model is accessible to

farm managers and their advisers. For a model to be used effectively,

it needs not only to mimic accurately the real system but also to be

accessible to managers. Effective communication between the model and

its intended users is most readily obtained by incorporating the model

into an information system which includes a comprehensive data

collection and transmission service.

A six—plant, three—firm system simulation model of the processing

sector of the domestic vegetable oil industry was constructed by

specifying the technical parameters of representative stage, plant, and

firm production functions item by item (Lamm 1976). Five measures or conceptual approaches were applied to market performance analysis when price and cost data were not available. These were productivity measures, marketing bill measures, flow analysis, market structure measures, and the application of welfare economics. Using the · 56

simulation approach allowed for the segregation of individual product

markets from aggregates and for the use of explicit measures of market

performance, measures that were more precise than the vague and

sometimes ~isleading market structure measures frequently used.

A stochastic simulation model was specified by Bigman and

Reutlinger (1979) to assess the impact of trade and buffer stock

policies on the stability of consumption and prices and the expected

values and standard deviations of costs and gains to consumers,

producers, and the government, and the balance of payments. Trade

policies were shown to have a greater impact on the stability of a

country's food grain supply than any reasonable size buffer stock.

APPLICABILITY OF EXISTING MODELS

The usefulness of plant location models as predictive devices for

determining the flow of pallets in the grocery distribution system

appears to be limited by the focus of these models on costs,

particularly the cost of transfer from supply point to demand point.

These models assume that firms will make their location and output

decisions in such a way as to minimize industry costs, based on known

levels of demand for the product. Bobst and Waananen (1968) maintain

that firms are much more likely to base these decisions on profit

maximizing criteria than on cost minimizing ones. Only in the case of a

perfectly competitive industry will the profit maximizing criteria and

cost minimizing criteria result in an identical location or output 57

decision. As these criteria deviate from one another, actual spatial

organizations will differ from those predicted by the models, and the

models will underestimate realized industry costs.

Therefore, I reject the plant location modeling approach in

modeling the demand for grocery pallets by the grocery distribution

industry on the basis that the grocery distribution industry is not a

perfectly competitive industry, and furthermore that regional quantity

of grocery pallets demanded is unknown.

Most spatial models, linear or nonlinear, and most supply-response

models analyze specific economic activities, such as the production or

the distribution of a commodity. Spatial models use a region as the

basic producing unit: supply-response models use an individual

(representative or typical) firm. Supply-response models try to predict

supply market under a given set of market conditions - ignoring, for the most part, interactions between firms in different regions or, for that

matter, between firms in the same region. Spatial models try to take

account of interregional competitive forces by explicitly including demand restraints and permitting interregional commodity shipments.

Insofar as they are inconsistent with regional effects, individual—firm effects are largely ignored (Heady and Hall 1968).

Both the linear and nonlinear models postulate a linearly homogeneous production function. To the extent that such a production function is a distortion of reality, it is an inherent weakness of these models. Heady and Hall (1968) note that supp1y—response models make the 58

same assumption. Spatial models also assume complete resource mobility

among firms in a given region. This assumption is obviously

inconsistent with short—run equilibrium. As more regions are

incorporated, the distortion is reduced. A distinct advantage of supply-response models is that, by their very nature, they include resource immobilities between firms.

The pattern of spatial prices is explained largely by transfer costs such as transport, marketing margins, and governmental trade barriers. Spatial equilibrium models have attempted to explain these price differentials by assuming that the products being traded are perceived by demanders as being homogeneous. These models, however, do not allow consumer prices of a commodity in a particular import market to vary by supplier at a given point in time. In addition, they predict a unidirectional trade flow pattern, so that either cross-hauling of commodities between two trading regions is prohibited or the trade pattern predicted is a net trade pattern. Commodities supplied by different regions may be differentiated in the eyes of importers.

If the pallet is considered to be a differentiated product, models based on product differentiation may be justified in at least two ways.

First, the demand theory of Lancaster (1966) states that consumers ultimately desire product characteristics, and products can combine these characteristics in varying proportions. Johnson, Grennes, and

Thursby (1979) outline an alternative approach which stresses differences among suppliers rather than product characteristics. Some . 59

suppliers may be more reliable than others and buyers may want to

diversify to protect against the possibility of supply interruption. If

suppliers offer different contracts in terms of delivery dates, credit,

or other services, prices of identical products in a market may vary by

supplier. If particular suppliers are more reliable than others, then a

contract delivered from a reliable supplier is of a different quality

than a contract made with less reliable suppliers.

Commodities are differentiated frequently by quality or place of

origin. Monke and Petzel (198ü) propose two complimentary tests -

bivariate price regressions and hedonic index estimation - as methods to

identify whether differentiated products are amenable to treatment as a

homogenous commodity. These price linkage tests represent necessary,

rather than sufficient, conditions for aggregation and must be

supplemented by information on market structure. The distinction

between homogenous and differentiated products has obvious importance

for the formulation of empirical models of regional trade. A homogenous

commodity obeys the law of one price, in which prices across regions can

differ by no more than the cost of commodity arbitrage. Analyses of

regional trade have often assumed differentiated products to imply

distinct markets without systematically testing for linkages. The

source of differentiation usually has involved the place of origin or destination or quality variation.

Although prices of different products may differ, this phenomenon does not preclude an aggregate treatment of these products. If 60

differentiated products demonstrate a high degree of substitutability in

production or consumption, the shocks from changes in the supply and

demand of one product are transmitted to other products in the commodity

group. In the case of pallets, the wood raw material input to the

production of grocery pallets can alternatively be used to produce

non-grocery pallets. The products are differentiated by end-use; but,

the raw material input has a high degree of substitutability. This

mechanism leads to price linkages across the differentiated products

that can be identified statistically. Integrated markets are defined as

markets in which prices of differentiated products do not behave

independently. Markets which are independent must be modeled in a

disaggregate manner, while markets which are integrated may be amenable

to aggregate analysis.

As noted earlier, spatial equilibrium models depend on the

identification of transfer costs for products being traded between

regions. Identification of regional transfer costs for grocery and

related products is complicated by the fact that reform or elimination

of most state and federal economic regulations of motor carriers

(Beilock and Freeman 198N) has coincided with the elimination of most

data collection by those agencies with regard to motor freight rates.

Thus, identification of regional transfer costs would require separate

surveys of private motor carriers, rail car shippers, and grocery distribution captive fleet operations. Although such surveys are considered to be outside the scope of this study, they could be 61

conducted to provide a point estimate of current transfer costs between

regions.

In addition, spatial equilibrium models are designed to predict a

unidirectional trade flow where cross-hauling of commodities between two

trading regions is not considered. If the objective of this study was

limited to modeling only the demand and supply of grocery products on a

regional basis, a spatial equilibrium model for the unidirectional flow

of grocery products from manufacturer to retail store would be

appropriate. However, the objective particularly involves modeling the

demand and supply of grocery pallets in the grocery distribution

industry. In modeling the demand and supply of grocery pallets, I must

consider that the movement or use of pallets in grocery distribution is

not unidirectional. That is, pallets continuously circulate through the

system until they are no longer serviceable. This continuous

circulation of pallets presents a complication which spatial equilibrium

models are not equipped to handle. For this reason, I also reject the

spatial equilibrium modeling approach in modeling the demand for grocery

pallets by the grocery distribution industry.

Econometric models which identify price-quantity relationships

using a system of demand and supply equations are limited in their

application to the analysis of the grocery distribution system for several reasons. First, these models depend on the availability of time series data which identifies the historical relationship between quantities demanded or supplied and price. Sufficient historical data 62

do not exist for the building of statistical models which would permit

such an analysis. Second, the demand for pallets in the grocery

distribution system is complicated by the existence of inventory levels

of pallets already in the system. The linkage between inventory levels

and the price of pallets demanded from outside sources has been

identified as a critical aspect of price determination. Again, the lack

of price data for grocery pallets prevents any statistical estimation of

the effects of this linkage. Finally, the supply of pallets to the

grocery distribution system cannot be simply estimated as a function

which relates the production and material costs to the quantity of new

pallets supplied. Used pallets, repaired pallets, and exchanged pallets

also enter the distribution system in addition to the new pallets

supplied. The problem here is not with the used pallets or the repaired

pallets which have prices associated with the quantities supplied to the

system, but rather with the exchanged pallets that have no price

associated with them. These exchanged pallets can affect the life

expectency of pallets in the system which in turn effects the rate of

replacement and ultimately effects the quantities supplied. However,

there is no price linkage which can be estimated between total

quantities supplied and the quantities of pallets exchanged.

One aspect of the econometric modeling approach included in this study is the identification, in descriptive rather than statistical terms, of the components which make up the various elements of demand, supply, and price determination in the associated market levels of the 63

distribution system. The variables which affect the demand for grocery

products are linked with those which affect the demand for movement of

grocery products and these, in turn, are linked with the variables which

effect the demand for pallets in the system. The price mechanism exists

which serves to clear the market, but it can only be described

schematically rather than in statistical terms.

The existence of inventory quantities of grocery pallets in the

grocery distribution industry requires a different econometric approach

to demand modeling for grocery pallets. The stock adjustment modeling

approach permits identification of the quantity of grocery pallets

demanded by expressing the quantity demanded as a function of expected

sales of grocery and related products. I will use this approach to

develop a model that first identifies the inventory quantity of pallets

in the system and subsequently relates inventory quantity to the

quantity of pallets demanded. This model can be used in assessing the

long-term potential and long-term trends in the grocery pallet market in terms of the level of retail sales in the grocery and related products industry.

Simulation modeling is the final approach considered for analysis of pallet use by the grocery and related products industry. One advantage of using a simulation approach is that it allows for the segregation of individual pallet markets rather than having to aggregate production or consumption over the entire country. Another advantage is 64

that the lack of price and cost data does not present a restriction on

model development as it would with other model forms.

Simulation techniques are designed to imitate a system's behavior

over time. Therefore, an analysis of the flows of grocery products on

pallets between market areas to achieve a known level of retail sales in

a given year is possible. However, the objective of this research is to develop a model for analysis of future demand and supply of grocery

pallets. Application of the simulation modeling approach would require

that future levels of retail grocery sales be estimated or predicted,

which is outside the capability of the simulation modeling approach.

Even if retail grocery sales were predicted using an alternate

econometric approach, application of simulation techniques to analyze

the future flows of grocery products on pallets would lack any measure of the error associated with the predicted flows of pallets. For these reasons, I also reject the simulation modeling approach in modeling the demand for grocery pallets by the grocery distribution industry. CHAPTER V

GROCERY DISTRIBUTION MODEL

The initial section of this chapter describes the spatial

relationship between various market levels of the grocery distribution

system. In the next section, I describe the movement of pallets between

grocery demand and supply points in order to identify the relationship between movement of grocery products and the demand for pallets at each point of shipment origin. Demand and supply relationships in the grocery distribution system are presented in the following sections of the chapter. Finally, a stock adjustment model for grocery distribution is presented.

SPATIAL ASPECTS OF GROCERY PRODUCTION AND DISTRIBUTION

Grocery production and distribution is not randomly distributed among the various regions of the country. Rather the location of these economic activities is determined by a number of factors, such as regional endowment of resources, production costs for intermediate and final products, transfer cost functions, and demand functions for the final products. The relative level of these factors determines the comparative advantage of one area to another. This, in turn, influences both the direction and extent of growth and development of the above activities in the region (Nichols 1969). Two examples are used to

65 66

illustrate this point. First, the 1982 Census of Manufactures ( U.S.

Dept. of Commerce 1985) reports a total of 16,813 companies

manufacturing food and kindred products (SIC Code 20). However, only 32

companies are reported manufacturing cereal breakfast foods (SIC Code

ZOM3). The eight largest of the 32 companies account for over 90

percent of the value of shipments, which are reported to exceed U.1

billion dollars in 1982. More importantly, the cereal breakfast food

companies located in the North Central Region account for over 70

percent of the value of shipments. The firms located in the North

Central Region have a comparative advantage over firms located in other

regions, in terms of resource availability, lower cost for raw material

inputs, and lower production costs resulting from economies of scale.

The comparative advantage for cereal breakfast food manufacturers in the

North Central Region results in a concentration of firms in the region.

As another example, frozen orange juice processing firms are concentrated in Florida and California. The 1982 Census of Manufactures reports value of shipments of frozen juice products exceeding 1.8 billion dollars in 1982, with approximately 79 percent of the value of shipments coming from Florida and California. Kilmer, et al (1983) notes that orange juice processing plants are typically located in close proximity to the orange groves. These firms minimize the cost of getting the raw material to the processing plant. Also, by shipping frozen juice rather than unprocessed juice oranges, the value of each pallet load of product shipped is increased. From our survey of 67

distribution centers, it is estimated that the average pallet load of

frozen food products is valued at $800 while produce, like juice oranges, is valued at less than $ü00. Thus, the comparative advantage over other regions in manufacturing and shipping frozen orange juice

results in these production activities being concentrated in the two states. _

Grocery production and distribution activities can be centered in reglons spatially separate from the location of demand for grocery products (Figure 1). Because of the tendency for food production, distribution, and consumption to be spatially separated activities in the food production chain, each stage in processing generates a demand for the movement of products. This demand is satisfied, particularly in the latter stages, by materials handling services that include the use of pallets.

Grocery Manufacturers

The raw materials for grocery and related products are moved from a farm or other intermediate manufacturing site to the final product manufacturing plant, symbolized in Figure 1 as an arrow between Raw

Material Producer and Grocery Manufacturer. An example of raw material movement is the shipment, by railcar, of grain products from grain elevators in the North Central Region to Kellog Co., in Battle Creek,

Mich., and Ralston Purina Co., in Cincinnati, Ohio, both manufacturers of cereal breakfast foods located in the North Central Region that 68

Distribution Center A

Pallet Manufacturer

fl·~ ^ ~ * Distribution Center B 'allet Manufacturer B

Distribution Grocery Center C Manufacturer

Pallet Manufacturer Raw Material C Producer

Figure 1. Spatial Aspects of Grocery Distribution. 69

distribute their products nationally. From the Grocery Manufacturer,

final products are moved to grocery distribution centers, symbolized in

Figure 1 as arrows between Grocery Manufacturer and Distribution Centers

A, B, and C. An example of this movement of final products is the

shipment of cereal products from Kellog Co. and Ralston Purina Co. to

Kroger distribution centers in Cincinnati, Ohio, and Roanoke, Va., and

to a cooperative distribution center, Richfood, Inc., located in

Richmond, Va.

Some grocery and related products manufacturers maintain regional warehouses, spatially separated from their manufacturing facility, from which products are supplied to grocery distribution centers. Examples of these include General Foods, Morton-Norwich, and Quaker Oats Company.

Thus, the shipment of palletized product from grocery manufacturer to grocery distribution center may involve a trans—shipment through a grocery manufacturer's regional warehouse. But, the shipment is originally palletized at the grocery manufacturing facility (the pallet demand point) and palletized product moves through the manufacturer's regional warehouse without being unloaded from the pallet. Therefore, the function describing the grocery manufacturer's supply of grocery and related products to grocery distribution centers requires the same number of pallets even though the palletized product may not proceed directly from the manufacturer to the grocery distribution center. 70

Grocery Distribution Centers

The preceeding example considers distribution centers operating in

two separate geographic areas, designated as the Cincinnati market area — and the Charleston Roanoke market area. The 1987 Progressive Grocer's

Marketing Guidebook reports that 18 distribution centers serve 3,üü5

retail grocery stores in the Charleston - Roanoke market area, and 17

distribution centers serve N,923 retail grocery stores in the Cincinnati

market area. These 35 distribution centers receive cereal breakfast

food shipments from both of the above manufacturers, one of which,

Kellog Co., is located outside the market area served by all 35

distribution centers, and the other, Ralston Purina Co., is outside the

market area served by the 18 distribution centers in the Charleston -

Roanoke market area.

Both demand and transfer cost functions for grocery products are

critical in determining the location of grocery distribution centers in

relation to the retail stores. The demand for grocery products must be

sufficient to justify the investment needed to operate a distribution

center. From values reported by Kaylin (1968), adjusted to reflect

current prices, it is estimated that a distribution center can be

operated profitably with the value of products shipped by the center as

low as 20 to 26 million dollars annually. Thus, the retail stores served by the distribution center must be expected to generate sales equal to or exceeding this range to justify establishing a distribution center in a particular location. However, the transfer cost function 71

places a limit on the area which can be profitably served by a

distribution center. Using Figure 1 as an example, the maximum distance

that a retail store can be located from a distribution center and still

be profitably served is indicated by the outer circle around each

distribution center. This distance varies as the transfer cost function

for each distribution center varies; however, Kaylin (1968) notes that

150 miles is the average limit on distance between distribution center

and retail store for maintenance of acceptable ratios of expense to

total volume of products transported. Industry sources indicate that

this distance is still valid in transportation of grocery products to

retail stores in the current market, particularly in the Eastern half on

the United States.

Shippers of grocery products must consider the full transport cost

and not just the freight rate. The full transport cost is the freight

rate plus any nonprice costs associated with the service quality offered

by the mode of transportation. These characteristics include speed,

reliability, flexibility regarding scheduling, routing, shipment size,

load handling and monitoring characteristics, and claims handling procedures (Beilock and Casavant 198N). The magnitude of the full transport cost is determined by the cost and availability of a dependable and flexible transportation system, the cost of labor and other materials handling techniques, and the distances between shippers and receivers. In order to minimize transport cost and maintain the flexibility required in grocery distribution to retail stores, most 72

distribution centers maintain their own fleet of trucks, hereafter

referred to as a captive fleet, which are used to deliver grocery

products to retail stores. However, products may be shipped to the

distribution center using common carrier trucks, captive fleet trucks,

or rail cars, depending on the cost efficiencies of the alternate

carriers.

‘ Retail Stores

The locations of retail stores served by each distribution center

are indicated in Figure 1 as a series of circles around each

distribution center. In reality, the distance of retail stores from the

distribution center is not uniform as indicated by the circles; however,

the circles serve to indicata the geographic area within which a

distribution center operates. Distribution Centers A and B serve some

retail stores located in the same geographic area, symbolized by the

overlapping circles. The Kroger distribution center in Roanoke, Va.,

and the Richfood distribution center in Richmond, Va., are examples of

this. A total of 79 supermarkets (retail grocery stores with annual sales of over 2 million dollars) are served by Kroger and 25 supermarkets are served by Richfood in the Charleston - Roanoke market area. The Kroger distribution center serves a total of 107 supermarkets, with 28 located outside the geographic area designated as the — Charleston Roanoke market area. The Richfood distribution center serves a total of 2ü1 supermarkets, with 216 located outside the 73

Charleston - Roanoke market area. This example further illustrates the

fact that few distribution centers serve retail stores in exactly the

same geographie area even though their service areas frequently

overlap. It is important to note that retail stores are typically

served by a single distribution center. Thus, even though retail stores

are located in the same geographie area, they may or may not be served

by the same distribution center. whether they are or are not served

depends primarily on the corporate affiliation between the retail store

and the distribution center.

The Kroger distribution center in Cincinnati, Ohio, is an example of a distribution center (C in Figure 1), which serves retail stores in an entirely separate geographie region. Although it is part of the same corporate chain as the Roanoke, Va., Kroger distribution center, the

Cincinnati, Ohio, Kroger distribution center serves 86 Kroger supermarkets solely in the Cincinnati market area and is not involved in distribution to stores in the Charleston - Roanoke market area.

Pallet Manufacturers

A pallet manufacturer supplies pallets to distribution centers and grocery manufacturers in response to their demand for materials handling services. In Figure 1, only Pallet Manufacturer A supplies pallets to more than one distribution center. Pallet Manufacturer C, on the other hand, is able to profitably supply pallets to both Distribution Center C and the Grocery Manufacturer. In general, pallet manufacturers are 7N

found to be located in close proximity to the purchaser of their

pallets, primarily because of the limits that the cost of transportation

place on the distribution of pallets. As mentioned earlier, the average

limit on the distance between pallet manufacturer and purchaser is 50

miles. The average distance between distribution centers within the

same corporate organization is approximately 300 miles however, based on

the fact that distribution centers may serve retail stores located 150

miles away. Therefore, it is likely that a pallet manufacturer serving

one distribution center in a corporate chain like Kroger would not serve

another distribution center in the same chain. If one considers that n the 1987 Progressive Grocer's Marketing Guidebook identifies NO9

distribution centers located throughout 55 geographic areas of the

country and that Emanuel (1985) identifies 2,ü7O pallet manufacturers,

also located throughout the various regions of the country, then the

probability of finding a pallet manufacturer within 50 miles of either a

distribution center or grocery manufacturer is very high. In fact,

Emanuel's tabulation of pallet manufacturers indicates that 17 pallet

manufacturers are within 50 miles of the Kroger distribution center and

Ralston Purina food manufacturer in Cincinnati, Ohio. His survey also

indicates that 11 pallet manufacturers are within 50 miles of the Kroger

distribution center in Roanoke, Va. and another 16 are located within 50

miles of the Richfood distribution center in Richmond, Va.

One must keep distinct the movement of pallets between pallet

demand and supply points, from the movement of pallets between grocery 75 .

demand and supply points. As identified above, there are potentially

2,ü70 pallet supply points located throughout the United States. The

1982 Census of Manufactures reports 22,130 establishments, including

multi-establishment companies, engaged in manufacturing food and kindred

products which may be combined with the MO9 distribution centers

identified above to yield potentially over 22,500 pallet demand points

in the grocery distribution industry.

With regard to the movement of pallets between grocery demand and

supply points, 108,600 retail grocery stores are reported in the 1987

Progressive Grocer's Marketing Guidebook, excluding convenience stores which operate outside the relevant grocery distribution system and

account for only 7 percent of total retail sales. If one counts the ü09 distribution centers as both grocery demand and supply points in terms of the movement of pallets between these points, then there are over

109,000 potential grocery demand points and over 22,500 grocery supply points where the movement of pallets occurs.

The movement of pallets between grocery demand and supply points does not represent trade in pallets but, rather, the utilization of pallets. The trade is in grocery items. As a result of grocery trade, pallets move from place to place, thereby influencing the net demand for pallets at a given point. It is the geographic movement of pallets in the course of their use that makes the estimate of regional pallet demand unique from, say, the demand for construction lumber where the demand for lumber at one point cannot be satisfied by the supply of 76

lumber to another point. The movement of food from place to place

influences the demand for pallets at each point of shipment origin,

which may be either a grocery manufacturer or a grocery distribution

center. If at each point of shipment origin the supply of pallets from

incoming cargo, including pallets involved in backhaul shipments, is

inadequate to handle outgoing cargo, the deficit must be met by new or

reconditioned pallets. In order to understand how pallets are supplied

at each point of shipment origin and thereby determine the demand for

new or reconditioned pallets, one must consider how pallets move through

the grocery distribution system.

PALLET FLOWS IN THE MODEL

Grocery pallets are durable goods in the sense that they provide a

flow of services, i.e., carrying grocery and related products, over a

period of time. Thus, a new pallet does not have to be purchased every

time a palletized load of grocery and related products is moved from a

manufacturer to a distribution center or from a distribution center to a

retail store. Generally the movement of grocery products is — unidirectional from the manufacturer of the grocery product to a

distribution center and from there to a retail store, as illustrated in

our previous example. The movement or use of pallets in the grocery distribution system is not unidirectional but is instead circular in nature - from the distribution center to the retail store and back to

the distribution center, for example. The demand for new pallets at the 77 distribution center to satisfy this circular flow is zero only if the number of pallets returned from the retail stores in usable condition

are equal to or greater than the number of pallets needed to move grocery and related products from the distribution center to the retail stores. The demand for new pallets at the distribution center is also zero in the case of movement of palletized product from a manufacturer to the distribution center that is coupled with the direct exchange of pallets at the distribution center.

Grocery Distribution Center

This analysis concentrates on the grocery distribution center as the focal point of activity with regard to the movement of grocery and related products and resulting flow of pallets (Figure 2). It is important to note that some pallet flows, indicated as dotted lines in

Figure 2, occur outside the distribution system surrounding an individual distribution center. These flows are important because they clearly indicate that the distribution center system is not a closed system. Pallets flow from pallet manufacturers and from grocery and related products manufacturers within the system diagramed to systems surrounding other distribution centers. In addition, other pallet manufacturers and other grocery products manufacturers, that are not involved in the flow of pallets within the system diagramed, also move pallets to other distribution centers outside the system diagramed.

Thus, the distribution center system is a component of a larger, closed 78

=•••¤•¤•••¤¤ ·...·.. E Grocery & Related Products E E Manufacturers I

Pallet g Distribution

G & R l t d P ducts Pallet Manufacturers roceräanufäcäuiersro I I I . | Retail Grocery Stores

Pallet Repair

FacilitiesDistribution Center

Discarded Pallets

Figure 2. Flow of Pallets in a Grocery Distribution System. 79

system that includes all distribution centers, pallet manufacturers, and

grocery and related product manufacturers.

A model of the grocery distribution system centered around the

distribution center is important because it shows the relationship

between quantity of new pallets entering the system and quantity of

grocery and related products moving through the system. Because pallets

are used to warehouse grocery and related products in the distribution

center, the model must also include those pallets in order to accurately

reflect the quantity of new pallets needed in the system. The total

quantity of pallets in use at a distribution center is equal to pallets

in the distribution center's warehouse area minus pallets leaving the

distribution center plus pallets coming into the distribution center.

These three pallet uses will be noted as "warehouse”, "outflow", and

"inflow".

Warehouse

The quantity of pallets in the distribution center warehouse is not static, but is constantly changing as the quantity of grocery products stored in the warehouse changes. Warehouse pallets may be divided into stock and float pallets. Stock pallets carry or support products in storage. Industry sources indicate that a large distribution center warehouse with over one million square feet of floor space typically has over 100,000 pallets in use as stock pallets. An average sized distribution center warehouse with approximately 500,000 80

square feet of floor space typically has over #0,000 pallets in use as

stock pallets. As grocery products are removed from storage daily for

shipment to retail stores, a portion of the stock pallets used to carry

those products in storage become available for other uses and become

additions to the quantity of float pallets. Float pallets are defined

as pallets that are not carrying or supporting products in storage at a

given point in time. That is, the float pallets are available for

storage of products not on pallets when they arrive at the center, for

exchange with pallets arriving from a grocery manufacturer, or for

movement of grocery products from the distribution center to the retail

stores. The quantity of float pallets in a typical distribution center

varies with the daily ~ovement of grocery products in and out of the distribution center but generally is less than one-tenth of the quantity of stock pallets in the warehouse.

Thus, at any given point in time, the total quantity of pallets at a distribution center (QDC) is composed of the quantity of stock pallets used for storage of grocery and related products and the quantity of float pallets used to absorb the fluctuations in the quantity of pallets moving in and out of the distribution center. This quantity may be represented as follows:

QDC = Stock + Float

While this equation includes all pallets in the distribution center, it does not include the flow of pallets moving through the distribution 81

system surrounding the center that result in pallets leaving the center

(outflow) and pallets returning to the center (inflow).

Pallet Outflow

The quantity of pallets in the outflow from the distribution

center includes pallets used to transport products to retail stores,

pallets returned to grocery manufacturers, pallets sent to repair

facilities, and discarded pallets (Figure 3). In the first two flows,

the quantity of pallets leaving the distribution center depends on the

demand for grocery product movement within the grocery distribution

system. In the latter two flows, the quantity of pallets depends on the

wear-out rate of pallets in the system and the frequency of damage to

pallets in the system. In the following sections, we examine each of

these flows in more detail.

Retail Store

Pallets carrying grocery products from distribution center to retail stores constitutes the single largest outflow of pallets from the distribution center. Using a large distribution center from the survey as an example, daily shipments to retail stores average over #700 pallet loads of products each day. This is based on a reported average of 1500 trailer shipments per week and 22 pallet loads of product in each trailer. An important point to note here is that these pallets reduce the quantity of float pallets available at the distribution center. 82

Pa]]€t Manufacturers Grocery & Related Products Manufacturers I I | Retail Grocery Stores

Pallet Repair Facilities

Distribution Center

Discarded Pallets

Figure 3. Pallet Outflow from a Distribution Center. 83

Grocery Manufacturer

Pallets are returned to grocery manufacturers as a result of prior

agreements with selected grocery manufacturers. Industry sources note

that approximately 50 percent of the grocery and related products

arriving each day come from manufacturers who have prior agreements with

the distribution center management that involves a one-for-one exchange

of equivalent pallets. That is, if the manufacturer ships a trailer

load containing 22 pallet loads of product to the distribution center,

the manufacturer expects the trailer to return to the manufacturing

facility with a similar number of unloaded pallets of quality equal to

the delivered pallets. Because the distribution center places the 22

pallet loads of products directly into storage, the pallets returned to

the manufacturer must come from the quantity of float pallets available

at the distribution center. For a large distribution center, this

direct exchange would amount to an outflow of about 2000 pallets per

day.

Pallet Repair

The quantity of pallets in the distribution center is also reduced when pallets are taken out of service for repair. The decision made by the distribution center manager to repair rather than replace damaged pallets with newly purchased pallets is based primarily on the difference between the cost of repair and replacement cost. The surveyed distribution center managers indicate that the cost of in—house 8l!

repair is limited to 25 to 30 percent of the cost of a new pallet. This

limitation results from space, time, and equipment constraints at

in-house repair facilities rather than a constraint on the cost of

repair. Storage space for pallets awaiting repair in-house is usually

limited because warehouse space is more profitably used in the storage

of grocery and related products. The limited storage space means that

pallets must be repaired and returned to service as quickly as possible

otherwise the pallets awaiting repair take up too much space. Finally,

the equipment generally available for in-house pallet repair is not

capable of the same level of dismantling and re-assembly of salvaged

pallet parts associated with outside pallet repair facilities. This

means that extreme rebuilding of a pallet is avoided and repairs

generally involve quick replacement of one to three deckboards.

However, there are substantial savings gained in repairing pallets

in-house over purchasing new pallets.

Although distribution center managers indicate that they consider

the expected life of repaired pallets to be equal to the expected life

of a new pallet, this observation must be considered to be biased as a

result of the managers' perception of the length of time that a pallet

remains in service in the distribution center. This perception is

related to the practice, described above, of direct exchange of S pallets. When a pallet leaves the distribution center as a result of a

direct exchange agreement, there is no guarantee that the same pallet

will ever return to the distribution center. Thus, the expected life of a repaired pallet is perceived by distribution center managers to be the

length of time that a pallet remains in a distribution center until it

is exchanged for another pallet. The exchange agreements only call for

an equivalent quality pallet to be returned. Because of the uncertainty

of getting back the same pallet in any exchange, distribution centers

prefer to exchange a lower cost repaired pallet as opposed to a higher

cost new pallet.

The outflow of pallets for repair constitutes a reduction in the

quantity of float pallets primarily because the pallets are sorted at

this stage in the warehouse operation. That is, empty pallets are

evaluated prior to being loaded. Warehouse personnel are instructed to

visually inspect each pallet before placing grocery products on it. If

a pallet is found to be damaged or no longer serviceable it is taken out

of the system for repair. With the exception of pallets which are so

badly damaged that they are beyond repair, a distribution center usually

sends damaged pallets to a repair facility, either in-house or an

outside contractor, without assessing the extent of damage. At this point, outside pallet repair facilities dismantle and salvage pallets which cannot be repaired economically. In—house pallet repair

facilities discard pallets which cannot be repaired economically in the sense that they relinquish ownership of these pallet to pallet salvage operators. The criteria for what constitutes a repairable pallet in-house are consistent across the surveyed distribution centers. The general rule-of—thumb used allows up to four deckboards and one 86

full-length stringer to be replaced in the repair of a damaged pallet.

If the damage to a pallet requires more than four deckboards or more

than one fu1l—length stringer for repair, the pallet is sold to pallet

salvage operators. The cost of repair varies depending on the amount of

damage, but a typical cost of in-house deck damage repair ranges between

$1.25 and $1.75 for the centers surveyed. For outside pallet repair

operations, the average cost of repair is over $2.00 and the amount of

damaged parts replaced may exceed four deckboards and one full-length

stringer.

Another aspect of pallet repair concerns the quantity of pallet

parts purchased from pallet manufacturers by the repair facility for use

in the repair of pallets (dashed line in Figure 3). The total amount of wood raw material going into the repair of pallets is determined by the number of pallets repaired and the amount of wood used in each pallet

for repair. The total amount of wood raw material needed for grocery and related products distribution therefore includes the amount needed for repair as well as that needed for the production of new pallets.

Pallet Discard

The total quantity of pallet material discarded includes that discarded at the distribution center as well as that discarded by the outside repair facility. In general, grocery pallets are not totally discarded. Only the broken or damaged parts are discarded during pallet repair or salvage. However, in this study, pallet discard from in-house 87

repair facilities is considered to be that quantity of pallets which

cannot be repaired economically in-house. Because ownership of these

pallets is transferred to pallet salvage operators, these pallets are

considered to have permanently left the distribution system and thus

need to be replaced. The quantity of pallets sold for salvage varies

greatly from one distribution center to another; but, based on the

information provided by distribution center managers, this quantity

averages 30 percent of pallets damaged. This means that 70 percent of

damaged pallets are recovered through repair operations as described in

the preceeding section. For a large distribution center, the quantity

of pallets sold for salvage may exceed 35,000 pallets over a period of

one year.

Later in this study, the simplifying assumption is made that the

total quantity of pallets discarded or sold for salvage is a constant

proportion of the existing quantity of pallets in the system. However,

this quantity may also be expressed as a function of the expected

service life of a grocery pallet, and the average age of pallets in the

system. As the average age of pallets in the system increases, or as it

approaches the expected service life, the portion of pallets discarded

will increase in a given time period. As new pallets are added to the

system, the average age of pallets in the system decreases and the portion of pallets discarded also decreases. Implicitly contained in

the above assumption is a second assumption regarding the expected

service life of pallets. That is, the specifications for new pallet 88

construction which influence expected service life are also assumed to

be constant.

Pallet Inflow

The quantity of pallets entering the distribution center includes

pallets returned from retail stores, pallets loaded with grocery

products shipped from a grocery manufacturer, pallets returned from

repair, and pallets received from pallet manufacturers (Figure #). In

the following sections, these flows are examined in more detail.

Retail Store ^

The quantity of pallets returned from retail stores constitutes

the largest daily flow of pallets into the distribution center. For a

large distribution center, this quantity incoming daily would amount to

3600 pallets. The pallets returned from retail stores are considered as

additions to float pallets because they are returned empty from the retail stores and are available for storage or further movement of grocery products. The quantity of pallets returned from retail stores depends on two factors: the demand for movement of grocery products to the retail stores and the quantity of grocery products backhauled from grocery manufacturers in captive fleet trucks to the distribution center. In the example above, over #700 pallets were used to satisfy the daily demand for movement of grocery products to retail stores. If all pallets used to satisfy this demand were returned directly to the 89

paffet Manufacturers Grocery & Related Products Manufacturers

Retail Grocery Stores

Pallet Repair Facilities

Distribution Center

Figure 4. Pallet Inflows to a Grocery Distribution Center. 90

distribution center, then on the average, there would be N700 pallets

returned daily from retail stores. The difference between the M700

pallets sent to retail stores and 3600 pallets returned from retail

stores results from the use of 1100 pallets for grocery product

backhauls. These backhauls are made using captive fleet trucks which

make delivery of grocery products to retail stores and proceed to

grocery manufacturers with empty pallets picked up at retail stores

(shown in Figure N as an arrow between retail stores and grocery

manufacturers). The empty pallets are exchanged at the manufacturer for

pallets loaded with grocery products which are delivered to the

distribution center. Industry sources state that between 35 and Mo

percent of grocery products nationally are backhauled from grocery

manufacturers in captive fleet trucks to the distribution center.

The net effect of backhauls in captive fleet trucks is a temporary

reduction in the quantity of float pallets at the distribution center.

Where backhaul arrangements exist, grocery products are moved to distribution centers on the pallets that were exchanged at the grocery and related products manufacturers for pallets which would normally return empty from retail stores. However, because these backhaul pallets arrive at the distribution center carrying products, they are added to the quantity of stock pallets in the warehouse rather than the quantity of float pallets as in the case of pallets returned to the distribution center directly from retail stores. 91

Grocery Manufacturer

The quantity of pallets received daily from grocery manufacturers,

including backhauls, is the second largest flow of pallets into the

distribution center. For a large distribution center, this quantity

would amount to 3100 pallets, incoming daily. The pallets received from

grocery manufacturers are considered as additions to stock pallets since

they are used in the warehouse to store the products shipped from the

manufacturers. This includes pallets arriving on captive fleet trucks

carrying backhauls as well as on common carriers and on trucks owned by

grocery manufacturers.

_ Pallet Repair

Another inflow of pallets into the distribution system occurs when

pallets are returned to service after repair. Repaired pallets are

considered as additions to float pallets because they are available for

storage or movement of products in the warehouse. The quantity of

pallets involved in this inflow depends on the rate of repair as well as

on the total number of pallets sent to repair. As noted above, about 70

percent of damaged pallets are recovered through repair operations. In

an average size distribution center, the quantity of pallets returned to the system after repair averages about 100 pallets per day. On the

other hand, in a large distribution center this quantity may exceed 750

pallets per day. For example, one industry source reported a total of

U5,000 pallets repaired over a three month period. If this rate of 92

repair was maintained throughout the year, the quantity of pallets

repaired would exceed the total quantity of pallets in the warehouse.

One reason explaining the magnitude of the above numbers is that the

pallets continually flow through the system rather than remaining

stationary in the warehouse. It is in the handling of grocery products on pallets that the damage to the pallet occurs which requires repair.

Pallet Manufacturer

While the repair of pallets is generally a continuous activity, occurring on a daily basis, the purchase of new pallets occurs at less frequent intervals. The decision to purchase new pallets is made by the managers at the distribution centers. Although these managers may have different titles, such as Traffic Manager, Warehouse Manager, Director of Distribution, Service Superintendent, etc., they all have the responsibility for purchase of new pallets to maintain the quantity of pallets available in the distribution system at some desired or optimum level.

The quantity of new pallets purchased by a distribution center is a function of two factors: the quantity of used pallets available in the system that includes the quantity of repaired pallets returned to the system and excludes the quantity of pallets discarded, and the volume of grocery and related products to be moved. Purchases are made by managers based on their knowledge of the quantity of pallets available in the system and their estimate of the quantity of pallets required to 93

handle the expected growth in volume of grocery and related products.

Industry sources report that pallet purchases for a large distribution

center can exceed 110,000 pallets per year. About one-third of the new

pallets purchased are replacement for pallets which are discarded or

sold for salvage because they are worn—out, damaged beyond repair, or

they do not meet the construction specifications required by the user.

The other two-thirds of the new pallets purchased are required to handle

growth in the movement of grocery and related products.

In order to balance the flow of pallets in a distribution system

between distribution center, retail stores, and grocery manufacturers,

new pallets must also be shown to enter the system through grocery

manufacturers (shown in Figure N as an arrow between pallet manufacturer

and grocery manufacturer). These new pallets may be purchased by a

grocery manufacturer or by a distribution center for delivery to the

grocery manufacturer as a result of an exchange agreement. For example,

Del Monte Corp., a grocery manufacturer in Swedesboro, N.J., has a

direct exchange agreement with a distribution center in Richmond, Va.

The distribution center receives products on pallets from Del Monte but

does not exchange pallets directly upon receipt of the product. The

distribution center purchases new pallets from a pallet manufacturer in

Philadelphia, Pa., in close proximity to the Del Monte plant. These pallets are delivered to Del Monte in exchange for the pallets delivered with product. Although it is noted above that distribution centers prefer to exchange used or repaired pallets rather than new pallets, in 9ü

this particular case the savings in transportation costs gained by

shipping the pallets directly to the manufacturer outweigh the cost of

new pallets delivered to the distribution center. Thus, total quantity

of new pallets entering a distribution system is the sum of new pallets

received by distribution centers and by grocery manufacturers.

DEMAND AND SUPPLY ASPECTS OF THE MODEL

In the following section, the demand and supply conditions are

described at two market levels, that is, the final product and pallet

market levels. In the ideal case where data limitations did not exist,

I would follow this description with statistical estimation, based on

times series data, of the relationships between quantity of grocery and

related products demanded and supplied and the quantity of pallets

demanded and supplied. In this case, however, I am limited to a

discussion of the theoretical relationships that exist in the final

product and pallet market levels because of the absence of times series

data on the factors involved.

As an intermediate product, pallets are not directly demanded by

the final consumer in the sense that grocery and related products are

demanded by the final consumer. However, the quantity of grocery and

related products supplied is shown to directly affect the quantity of

pallets demanded by the grocery distribution system. It is necessary to model both market levels because of the linkage that exists between the

final consumer demand for grocery and related products and the derived 95

demand for pallets as a means of moving the grocery and related products

to the point where the final consumer demand can be satisfied. A

diagram which illustrates the market levels also shows the relationship

between product supply and input demand at each market level (Figure 5).

As the equilibrium price is reached in the market for grocery and

related products, the demand for grocery and related products equals the

supply of those products. The price of grocery and related products is

a factor in the demand function for grocery pallets. Likewise, as the

equilibrium price is reached in the market for grocery pallets, the

demand for grocery pallets in the distribution and storage of grocery

and related products equals the supply of grocery pallets. Because

grocery pallets are inputs to the production function for distribution

and storage of grocery and related products, the price of grocery

pallets is a factor in the supply function for grocery and related

products.

Final Product Market Level

The final product market level for grocery and related products is

defined as the market for grocery and related products at retail grocery

stores. At this level, the demand for grocery and related products is a

final consumer demand determined by the weighted price of a market basket of grocery and related products measured in dollars per ton of product, the price of other goods and services, and consumer average 96

Demand A

Market A for Equilibrium E Price grocery products for A I Supply A ·<------———-——---I I Demand B ·<——---————-—--——I—----- Marke: B | for Equilibrium Price grocery pallets for B

Supply B

Figure 5. Demand and Supply Linkage Between Market Levels. 97

income. The quantity of grocery and related products demanded at retail

stores may be represented as follows: °D¤1> ’ f { Pops' Ps• Pc' Po• Ic } QDGP where is the total quantity of grocery and related products demanded by consumers measured in tons or cubic foot volume, PGPS is the price per ton or per cubic foot volume of grocery and related products at the retail stores, PS is the price of substitutes, PC

is the price of complements, Po is the price of all other goods and

services, and IC is average consumer income. The above functional

relationship should be considered as a stylized or classic

representation of the demand function for grocery and related products.

This relationship can be constructed to reflect the demand for grocery

and related products on a per-capita basis. In empirical specification and estimation of the grocery demand function, other variables may need

to be considered. These variables could include population demographics like age structure which may vary from one region to another and measures of the consumer price index in relation to the price of grocery and related products in different regions.

Substitutes for the quantity of grocery and related products demanded at retail stores could include other food sources such as convenience stores, farmers markets, and home gardens. Convenience stores are included as possible substitutes for demand at retail stores because the distribution system for convenience stores is considered to be separate from the distribution system for retail stores. That is, 98

the distribution system for convenience stores does not employ grocery

pallets in moving grocery products to the convenience stores.

Complements for quantity of grocery and related products demanded at

retail stores could include such factors as the price of cooking

utensils, opportunity costs associated with home food preparation, and

transportation costs incurred by consumers in obtaining grocery and

related products.

Given the above aggregate demand function faced by grocery

manufacturers, the supply function for grocery and related products

manufacturers includes the cost of raw material inputs as well as other

inputs to the production and distribution of grocery products. The

supply function faced by the grocery and related products manufacturer

may therefore be expressed as follows: Qscp ‘ f { P61>1>• Pmv Pnwpo } where is the total OSGP quantity of grocery and related products

supplied, is the price received by PGPP the manufacturer or producer of grocery and related products, PMH is the price of materials handling services, is the price of raw material PRM inputs used in the production of grocery and related products, and PO is the price of other inputs to the grocery production activities.

Two points must be noted concerning the above demand and supply functions. First, the price of grocery and related products at retail stores, includes the manufacturers or wholesale PGPS, price, PGPP, so that changes in the latter price are reflected in the former. The 99

reason for making a distinction between the retail and wholesale price

in grocery distribution is that the supply of grocery products goes

through multi-echelon inventory supply points with inventories

maintained at both the wholesale and retail levels. Because of this,

the grocery distribution center is both a demand and supply point for

grocery and related products. Thus, both the wholesale price and the

retail price are factors in the demand and supply functions for a

distribution center, as will be shown in the following section.

The second point concerns the price of materials handling

services, PMH. This is not a single value but rather is a weighted

value of all materials handling services incurred in moving grocery and

related products to the final consumer. Specifying a realistic weighted

value that could be used to estimate the above supply function requires

that the individual elements making up that value be identified.

Materials Handling Elements of Final Product Supply

Because of the spatial separation between producing and consuming points, as documented earlier in this chapter, materials handling services are needed to transport grocery and related products from their point of production to consuming centers. As noted above, the cost of materials handling services is included as a part of the production cost for the grocery products. This appears reasonable if the process of transporting a grocery product from producer to consumer is considered 100

to be a part of the process of preparing the grocery product for final

consumption.

Materials handling services for grocery and related products are

made up of several parts, including the transportation of grocery products fro~ manufacturers to distribution centers, the handling of products within distribution centers, and the transportation of products

from the distribution centers to the retail stores. It should be noted

that an open market exists for the transportation of grocery products

from manufacturer to distribution centers. Alternate forms of transportation in this market include captive fleet trucks owned by manufacturers or distribution centers, common carrier trucks, and rail cars. The choice of transportation mode must be made based on the least cost incurred in moving the grocery products from manufacturer to distribution center.

Materials handling services therefore include a number of elements which can be considered separately but are not necessarily independent.

The use of railcars and trucks to move products are examples. Railcars may be used in conjunction with trucks to move grocery products to distribution centers. However, almost all shipments to retail stores are accomplished by truck and, at this stage in the distribution system, truck and railcar shipping are not substitutable. Materials handling also includes automated conveyor systems, pallet handling equipment, pallets, and finally, labor. 101

Grocery Manufacturers

Having noted that the movement of products frequently occurs

between manufacturers in one region of the country and distribution

centers in a spatially separate region, I must consider that

transportation services needed for shipment of grocery and related

products are one of the inputs to grocery manufacturers production

functions. The cost of transportation therefore is a factor in

determining price and quantity of grocery and related products supplied

and changes in the cost of transportation will result in changes in the

price of grocery and related products. The functional relationship for quantity of grocery and related products supplied by grocery manufacturers in response to the demand for those products in different regions therefore includes the cost of transport and may be expressed as follows: QS =r{1> ,1>,p,p,1>} GP,MFG GPP RM MH P 0 where is total quantity of QSGP•MFG grocery and related products supplied by a grocery manufacturer for which materials handling services are required, is the price of PGPP grocery and related products that reflects the manufacturers production and distribution costs, is PRM the price of raw material inputs used in the production of grocery and related products, is now the unit cost PMH of transport measured in dollars per ton-mile or dollars per cubic foot volume-mile, PP is the weighted price of grocery pallets used in distribution, and PO is the weighted price of other factor inputs to the production function. This 102

is the same functional relationship expressed for the supply of grocery

and related products by grocery manufacturers in the preceeding

section. However, in this function the value of reflects only the PMH

unit cost of transport. Also, the function shows the price of pallets

separately from the weighted price of other factor inputs to indicate ‘ that pallets are a necessary input to the grocery manufacturers

production and distribution function.

Grocery Distribution Centers

Materials handling services are required at grocery distribution

centers for handling of grocery and related products in storage and for

transport of products to retail stores. These are the other two parts

of the materials handling services for grocery and related products

listed earlier. The first of these parts is specified as an inventory

problem where it is necessary to stock grocery and related products for

the purpose of satisfying the demand for those products at retail stores

over a specified time period. The economic parameters which determine

the quantity of grocery and related products handled in storage at the

distribution center may be expressed as follows: °D¤p,¤c ‘ “c· f { Sc' Papa Pops' PP } where is the quantity of QDGP,DC grocery and related products

handled in storage at the distribution center, SC is a setup cost

which is associated with the placement of an order to the manufacturer,

is the purchase price or PGPP production and transportation cost for 103

the manufacturer of the grocery products, is the price PGPS charged at

retail stores for the product, HC is a holding cost or the cost of

carrying the product in inventory, and PP is the price of grocery

pallets required to hold the products in storage. This function has

added importance in that the quantity of QDGP,DC, grocery and

related products handled in storage, is also the quantity of grocery and

related products demanded by the distribution center from grocery

manufacturers. This demand is derived from the distribution centers

production function which seeks to satisfy the demand for grocery and

related products at retail stores.

The second part is specified as a transportation distribution

problem which involves the determination of a minimum cost shipping

schedule to satisfy the demand for grocery products at retail stores at

several destinations with available supply. In this problem, the quantity of grocery and related products supplied by a distribution center equals the quantity delivered to retail stores. However, grocery distribution centers typically minimize the cost of transporting these products by applying linear programming techniques which consider the quantity of grocery and related products to be transported, and PMR, the unit cost of transport. Pallet use in transport of grocery products to retail stores is independent of distance traveled. Pallet use is a function of volume of products moved and demand for pallets is exogenous to the determination of minimum cost shipping schedules. 10lI

The functional relationship which identifies the volume or

quantity of grocery and related products supplied by distribution

centers in response to the demand for those products at retail stores

may be expressed as followsz ” QSGP,DC f { P¤1>P• PGPS' PP' Po } where is total quantity QSGP,DC of grocery and related products

supplied to retail stores by distribution centers, is the PGPP

purchase price of grocery and related products, is the price of PGPS

grocery and related products at retail stores, PP is the weighted

price of grocery pallets used in distribution, and PO is the weighted

price of other factor inputs to the production function. A weighted

price for other factor inputs is included because the production

function for a distribution center is satisfied from the use of other

materials handling factors as conveyors, pallet handling equipment, and

labor.

Demand for Materials Handling Services

The demand for materials handling services employing pallets is

the last materials handling element of final product supply that is

considered; but, it is the most important in terms of defining grocery pallet consumption. The demand for materials handling services employing pallets in the grocery distribution system is derived from the movement and storage of grocery and related products by grocery manufacturers and grocery distribution centers as outlined in the 105

preceeding two sections. This demand is considered to be a function of

the price of capital or the interest rate, the price of materials

handling equipment, the price of labor, and the price of alternate

systems. More importantly, this demand is a result of a past decision

by grocery distribution industry management to invest capital in a

system using pallets. Once this system was put into operation, the

quantity of pallets needed to operate the system is directly

proportional to the quantity of grocery and related products moved and

stored in the system. Pallets are input in terms of a fixed proportion

production function for materials handling services at both the grocery

manufacturers and the grocery distribution centers. That is, each ton

or cubic foot volume of grocery and related products supplied by

manufacturers or moved and stored by grocery distribution centers

requires a fixed quantity of pallets to satisfy the demand for materials handling services.

Because the materials handling services production function is of fixed proportions with regard to pallet input, the consumption function for grocery pallets can be expressed in terms of the quantity of grocery and related products moved and stored. In this functional relationship, the aggregate quantity of pallets consumed is not expressed in terms of the price of pallets, price of grocery and related products, and price of other factor inputs to the production function for distribution of grocery and related products. Instead, these prices are embodied in the functional relationships expressed for the quantities of grocery and 106

related products included in the consumption relationship. Thus,

changes in these prices can be expected to affect the aggregate quantity

of pallets consumed, although they do not appear directly in the

consumption function. The aggregate consumption of pallets in grocery

distribution is therefore represented as followsz QCP ” ‘ ‘ a1QSGP,MFG a2QDGP,DC a3QSGP,DC QCP where is total quantity of grocery pallets consumed in the materials handling of grocery and related products, QSGP,MFG is the total quantity of grocery and related products transported from grocery manufacturers to grocery distribution centers, QDGP’DC is the total quantity of grocery and related products handled in storage at grocery

distribution centers, is the total quantity of QSGP,DC grocery and related products transported from distribution centers to retail stores,

and al, az, and a3 are constants from the fixed proportion

production function for materials handling services which define the

quantity of pallets required to move or store a unit of grocery and

related products.

It is important to note that the constants al, az, and a3,

are not necessarily equal. Shipments of grocery products from manufacturers to distribution centers, related to the al constant, are

. exclusively unit—loads. That is, a given quantity of a single product

is loaded on a pallet for delivery to the distribution center. On the

other hand, shipments of grocery products from distribution center to

retail stores, related to the a3 constant, are normally mixed-loads, 107

with a variety of products in less-than unit-load quantity loaded on

each pallet. While the grocery and related products are stored on

pallets in the distribution center warehouse in unit-load quantities,

the constant az must also consider the storage of products which do

not arrive at the distribution center on pallets.

The aggregate grocery pallet consumption relationship including the constants al, a2, and a3, is a short-term relationship. When

~ the demand for materials handling services employing pallets changes as

a result of changes in the previously defined factors affecting that demand function, the constants al, az, and a3 will also be

changed. Thus, U in the long term, these constants can be altered by

changes in the price of capital, price of labor, price of materials

handling equipment, and price of alternate systems which may not use

pallets in the same proportion. As alternate systems are employed in

moving grocery and related products, the ratio of quantity of pallets

used to quantity of grocery products moved can change over time. It is

important to note that, in the short term, the constants that define the

quantity of grocery product which may be moved or stored on a pallet are

independent of the price of the pallet. That is, regardless of the

price of the pallet, the quantity carried on the pallet is limited and

defined by the constant. 108

Pallet Market Level

Demand for pallets in grocery distribution is derived from the

transportation and storage of grocery and related products by grocery

manufacturers and grocery distribution centers. In the preceeding

sections, the production functions and corresponding demand and supply

functions for grocery manufacturers and grocery distribution centers

have been identified. The demand for pallets in grocery distribution is

derived from these productions functions and is expressed in terms of

the price of pallets, the wholesale and retail prices of grocery and

related products, and price of other factor inputs to the production

functions for grocery manufacturers and grocery distribution centers.

The aggregate demand for pallets in grocery distribution is therefore

represented as follows: °D1> “ f { P1>• PGPP° Pops' Po } QDP where is total quantity of grocery pallets demanded for materials handling of grocery and related products, PP is the weighted

price of grocery pallets used in the distribution of grocery and related products, is the grocery manufacturers delivered PGPP price of grocery and related products, is the price of grocery and related PGPS products charged to retail stores, and PO is the price of other factor inputs to the production and distribution of grocery and related products. These other factor inputs include complements of pallet demand such as automated conveyor systems and pallet handling equipment, and substitutes for pallet demand such as manual labor and metal carts. 109

The quantity of pallets estimated by the aggregate pallet demand

relationship is the total quantity of pallets used in the movement and

storage of grocery and related products in a specified time period. The

total pallet demand relationship reflects changes in the volume of

grocery and related products moved or stored through changes in the

wholesale and retail prices of the grocery and related products.

The total quantity of pallets supplied in grocery distribution is

expressed as a function of the price of factor inputs to the production

function for grocery pallets. That is, QSP = f { PP' Po } PCAP° PLAB’ OSP where is total quantity of grocery pallets supplied for

materials handling of grocery and related products, PP is the weighted price of grocery pallets used in materials handling of grocery and related products, is the price of capital PCAP or the interest rate, PLAB is the price of labor, and PO is the price of other factor inputs to the production function for grocery pallets.

When the system is in equilibrium, the quantity of grocery pallets demanded and the quantity supplied are equal as shown by the following identity: ¤"P : ¤Sp ODP where is total quantity of grocery pallets demanded for QSP materials handling of grocery and related products and is total quantity of grocery pallets supplied for materials handling of grocery and related products . 110

Because grocery pallets remain in the system from one time period

to another, the total quantity of grocery pallets supplied for the

materials handling of grocery and related products includes new grocery pallets and grocery pallets already available in the system. That is, QS? “ ’ QNP Qmv QSP where is the total quantity of grocery pallets supplied for the distribution of grocery and related products, QNP is the quantity of new grocery pallets, and is the inventory quantity QINV of grocery pallets available at all supply points in the system. From this expression, we can identify the quantity of new grocery pallets demanded.

As noted earlier, the demand for new grocery pallets depends on two factors: quantity of grocery pallets available in the system and changes in the volume of grocery and related products moved. QINV QSP identifies the quantity of pallets available in the system, and identifies the quantity of pallets supplied to move a given volume of grocery and related products in a specified time period. Therefore, the demand for new pallets is the total quantity of pallets supplied less the available inventory of pallets in the system. That is, new pallet demand is an "excess demand" function, which may be expressed as follows: ¤°„P = QS. - ¤„„ QSP where is the quantity of new pallets demanded, QDNP is the total quantity of pallets supplied, and is the inventory quantity QINV O 111

of pallets available at all supply points in the system. Thus the

demand for new grocery pallets is a residual, or the difference between

total grocery pallet supply and the quantity of grocery pallets

available in inventory within the system. W

The total quantity of new grocery pallets supplied for the

materials handling of grocery and related products may be expressed as

follows: QS =r{P ,p,1>} NP NP RM O where is the quantity of new pallets QSNP supplied for materials

handling of grocery and related products, is the price of new PNP grocery pallets, is the price of raw material PRM inputs used in the

production of new grocery pallets, and Po is the price of other factor

inputs to the pallet production function. The quantity is the QSNP

total quantity of pallets produced by pallet manufacturers for all

grocery products manufacturers and distribution centers in the grocery

distribution system. The above new grocery pallet supply function is

the market supply function which, in equilibrium with the demand

function for new grocery pallets, determines the price and quantity of

new grocery pallets supplied to the grocery distribution system.

Other factors can also affect the price and quantity of new

grocery pallets supplied. In the short term, an increase in the demand

for non·grocery pallets could result in a price increase for grocery pallets. That is, pallet manufacturers would experience increasing marginal costs of production in producing an increased quantity of 112

non-grocery pallets. This would put upward pressure on the price of

pallets supplied, including grocery pallets. For those pallet

manufacturers operating at or close to maximum capacity, the galn from

produclng an additional, higher-value, non-grocery pallet would be

greater than that for produclng an additional grocery pallet; so, fewer

grocery pallets would be supplied. This again would put upward pressure

on the price of grocery pallets and as this price increased, both the

demand for pallet repair and the extent of damage which would be

repalred would tend to increase.

The quantlty of pallets available in lnventory includes those

available at grocery manufacturers as well as at the distribution A

center. This quantlty has an associated decay function which expresses

the quantlty available ln the system as a function of the quantlty

available in the preceeding time period less the quantlty expected to be

dlscarded. The quantlty of pallets in lnventory may be expressed as

follows: Qmv ’ Qt-1 ' eQt-1 " where OINV ls the total quantlty of pallets available ln lnventory for

materials handling of grocery and related products ln a given tlme

Qt_1 period, ls the quantlty of pallets ln the preceeding time w*Qt_1 period, and ls the quantlty of pallets dlscarded in the

preceeding time period. This decay function conslders the stock of

pallets available at all supply points ln the system and adjusts the 113

quantity available to reflect only those pallets available for the

movement of grocery and related products.

STOCK ADJUSTMENT MODEL FOR GROCERY DISTRIBUTION

The total quantity of pallets in use throughout the distribution

system constitutes the inventory of pallets in the system. Estimation

of the quantity of pallets in inventory is necessary for solution of the

pallet demand and supply relationships outlined in the preceeding

section. Recalling that pallets are a durable good, an application of a

stock adjustment model permits identification of the quantities of

pallets in inventory in the grocery distribution system. The stock

adjustment model also provides another approach to demand specification

for new pallets which emphasizes the importance of inventory levels to

the overall demand for new pallets.

Grocery distribution centers and grocery products manufacturers purchase new pallets in order to maintain a sufficient quantity of pallets for the materials handling of grocery and related products.

These purchases are based on the expected volume of grocery and related products passing through the distribution system. That is, purchases of new pallets are made in order to maintain a ”desired" quantity of pallets in inventory and this quantity is based on the expectations of grocery distribution managers regarding volume of grocery and related products that will require pallets for movement and storage. The

"actual" quantity of pallets available in the system, which involves a llü

physical count of all pallets in use throughout the system, constitutes

the inventory of pallets in the system. Considering that all of the

used and repaired pallets in the system were, at one time in the past,

purchased as new pallets, then this inventory quantity can be expressed

as a function of all past purchases of new pallets, or ’ Qt f {°m>,t· QNP,t-1° o'NP,t-2° °m>,u—¤ } where Qt is the actual quantity of pallets available in the system in

time period t, and is the quantity of new QNP’t pallets demanded in

time period t.

The inventory quantity of pallets maintained in a distribution system can also be expressed as a function of the dollar volume of grocery and related products sales. In a stock adjustment model, the desired quantity of pallets at the end of period t, or Q*t, and the dollar volume of grocery and related products sales during period t, or Xt, are related as follows: I Q f { Xt} t =

Although the quantity of pallets is changed by current decisions about the desired quantity of pallets, these decisions depend on all past quantities of pallets available with more emphasis on recent available quantities and corresponding less emphasis on quantities available long in the past. The adjustment made by the distribution system to the desired quantity of pallets in any one period may be expressed as followsz e u Ot·Ot_1 =8(Ot·Ot_1) 115

Qt where equals the actual quantity of pallets in the system at the

end of period t. The multiplier, g', represents the percentage

adjustment made between the actual and desired quantity of pallets in

any one period. In this study, the assumption is made that the time

period between the observations (1 year) is long enough so that the

actual quantity of pallets in the system is equal to the desired

quantity in any one period. Therefore, in this study the multiplier

equals one. Thus, for pallets in grocery distribution, 0t=¤°t=r{xt}

Qt where is the actual quantity of pallets in the system in time

t, Q*t period is the desired quantity of pallets in time period t,

and Xt is the dollar volume of grocery and related products sales in

time period t.

In the stock adjustment model, the net change in quantity of

pallets from one time period to another, Nt, is expressed as a

function of the changing dollar volume of grocery and related products

sales from one time period to another, or

=f{xt-xt—1} Nt=Qt-Qt—1

If the new pallets purchased each year remained in the system

indefinitely, then the net change in inventory quantity would represent

the total demand for new pallets in the grocery distribution system, or

In this case, if the dollar volume of grocery and related products sales decreased from one year to the next, there would be a negative demand 116

for new pallets. However, as noted earlier, the life of a pallet in a

grocery distribution system is finite. Therefore, one must also account

for replacement of discarded pallets which are lost because of excessive

damage and for the repair of damaged pallets which can be returned to

service after repair. Considering the replacement aspect first, the

assumption is made that the quantity of discarded pallets can be

expressed as a constant proportion of the quantity of pallets in the

system in the preceeding time period, or G wu ” (Qt-1) . " where Wt is the quantity of discarded pallets which must be replaced

in time period t, w· is the percentage of pallets previously purchased

that is affected, and is the quantity of pallets in the Qt_1 system in

the preceeding time period. A further assumption is made that a pallet can incur damage which results in the pallet being discarded at any time after the pallet enters the distribution system. In this assumption, the quantity of pallets being discarded is considered to be a constant function of the quantity of pallets exposed to potential damage. This assumption may be restated as one which assumes that replacement demand for discarded pallets is proportional to existing inventory. This is the same as an assumption of a declining balance depreciation formula used in other studies of stock adjustment, particularly Griliches

(1960).

The quantity of discarded pallets impacts on the relationship for net change in the quantities of pallets in the system as follows: 117

Nt ’ Qt ' Qt—1 ‘ wc

In this relationship, the quantity of pallets discarded further

increases the net change in the inventory quantity of pallets in the

system and thereby increases demand for new pallets in time period t. However, the introduction of the term Wt means that we cannot express

the net change solely as a function of dollar volume of sales without

some additional manipulation of the equation elements.

Substituting for Wt, ’ ‘ "*‘°t-1) Nt °t Qt-1 * Combining terms yields,

·I· Nt = Qt - (1 - w )Qt_1

This relationship may finally be expressed as a function of changes in

the dollar volume of retail sales I} for grocery and related products, or ut = r ( xt — (1 — w°}xt_1

In the case of repaired pallets, consideration must be given to

whether the repair is accomplished within the same time period that the damage occurs. If the quantity of pallets in a given time period equals

Qt and Rt pallets are removed from service for repair, the quantity of pallets in the system has decreased by the amount Rt. However, if the assumption is made that Rt pallets are repaired and returned to the system in the same time period, then the net change in the quantity of pallets is zero and the quantity of pallets in the system remains the same as before, or, Qt - Rt + Rt = Qt 118

Recalling that the repair of pallets occurs on a more or less continuous

(daily) basis and further that this study depends on yearly observations

of levels of retail sales, I conclude that the time period between

observations is sufficient to make this assumption viable.

One apparent limitation of this model is that it does not

explicitly consider the cost of repair in comparison to the price of new

pallets. The stock adjustment model deals only with the inventory

quantity of pallets in the system as a function of the dollar volume of

grocery and related products sales in the estimation of demand for new

pallets. However, I have noted that only the quantity of pallets

available in the system, along with the volume of products moved, influences the demand for new pallets. Then, as long as the repaired pallets are available, the cost of repair does not directly influence the quantity of new pallets demanded.

If a pallet is removed from the system for in—house repair and is found to have damage that exceeds the criteria previously enumerated for economic repair, that pallet is discarded, or as defined earlier, sold for salvage. The net effect is a reduction in the inventory quantity of pallets in the system. However, having made the assumption that the quantity of pallets sold for salvage is a constant proportion of the existing inventory, I can say that this quantity of pallets is already included as discarded pallets. Therefore, the constant, w*, in the net change equation, also accounts for those pallets which are removed 119

from the system for in-house repair but are lost permanently when it is

discovered that they cannot be economically repaired.

Thus, the net change in the quantity of pallets in the system,

after adjusting for the effect of pallets that have been discarded, does

constitute the quantity of new pallets demanded in the grocery

distrlbution system in a given time period. This quantity, noted earlier in this section as QNP,t, is expressed as a function of changes in the dollar volume of retail sales for grocery and related products: f‘ w*)Xt_1 (1) QNPJ = { Xt · (1 · } or, in terms of the quantity of pallets in the system: ” ‘°*)°t-1 (2) QN'P,t = Qt ' (1

At the beginning of this section, a functional relationship between the total quantity of pallets in use and past purchases of new pallets is given. Using formula (2) above, a mathematical expression

(developed in Appendix B.) is presented for the total inventory quantity of pallets in the system in terms of all past purchases of new pallets, or Q — N 2 = QNP,t+ (1 W (1 W ) + QINV )QNP,t-1+ - QNP,t—2 '°' Q (1 W - )nQNP,t—n

Although this expression considers all past purchases of new pallets, in practical terms one does not need to go any further back in time than the expected service life of an average pallet. For most grocery pallets, this time period will vary depending on the number of trips 120

that the pallet makes. For example, if the assumption is made that the

average life of a pallet in the grocery distribution system is two

years, in effect the assumption is that the annual replacement rate, w*, must be greater than or equal to 50 percent. At least half of the

pallets in the system are assumed to be replaced each year so that over

a two year period 100 percent of the pallets can be expected to be

replaced in the system.

Using the relationship in formula (1) above, the total inventory

quantity of pallets is finally expressed as a function of changes in the dollar volume of retail sales of grocery and related products: ‘ ‘1 ‘ °1¤v = f Xt **)%-1 * * Estimation of this functional relationship provides the second element in the ”excess demand" function noted in the preceeding section on the pallet ~arket level. 121

THE COMPLETE MODEL

The following demand and supply functions from the final product

market for grocery and related products and the market for pallets

illustrate the linkages between each demand and supply point in the

grocery distribution system and the price linkage between the quantity

of grocery and related products requiring materials handling services

and the resulting aggregate demand for pallets in the grocery

distribution system. It is important to note that the functions

representing the grocery pallet market could be estimated separately

from the functions representing the grocery and related products

market. That is, in future empirical estimation, the portion of the

model representing the grocery pallet market could stand alone and be

estimated separately from the demand and supply functions for grocery and related products.

Definitions of variables included in the functional relationships are summarized as follows: QDGP = total quantity of grocery and related products demanded by

consumers measured in tons or cubic foot volume.

retail price of PGPS = grocery and related products at retail stores.

PS = price of substitutes.

PC = price of complements.

PO = price of all other goods and services.

IC = average consumer income. 122

= total quantity of QSGP•DC grocery and related products supplied to

retail stores by distribution centers.

wholesale PGPP = price of grocery and related products.

retail price PGPS = of grocery and related products.

PP = weighted price of grocery pallets used in distribution.

PO = weighted price of other factor inputs to production function.

quantity QDGP'DC = of grocery and related products handled in

storage at the distribution center.

SC = setup cost associated with placement of order to manufacturer.

= wholesale price PGPP of grocery and related products.

= retail price of grocery PGPS and related products.

HC = holding cost or cost of carrying product in inventory.

PP = weighted price of grocery pallets.

= total quantity QSGP’MFG of grocery and related products supplied

by a grocery manufacturer.

PGPP = wholesale price of grocery and related products.

= price of raw PRM material inputs used in the production of grocery

and related products.

= unit cost of transport PMH measured in dollars per ton—mile or

dollars per cubic foot volume-mile.

PP = weighted price of grocery pallets.

PO = weighted price of other factor inputs to the production

function. 123

ODP = total quantity of grocery pallets demanded for materials

handling of grocery and related products.

PP = weighted price of grocery pallets.

wholesale price of PGPP = grocery and related products.

retail price of PGPS = grocery and related products.

PO = price of other factor inputs to production and distribution of

grocery and related products. OSP = total quantity of grocery pallets supplied for materials

handling of grocery and related products.

PP = weighted price of grocery pallets.

price of capital PCAP = or the interest rate.

price of labor. PLAB =

Po = price of other factor inputs to production function for grocery

pallets. 1 = quantity of new QDNP pallets demanded for materials handling of

grocery and related products. QSP = total quantity of pallets supplied.

= inventory quantity OINV of pallets available in the system.

quantity of OSNP = new pallets supplied for materials handling of

grocery and related products.

= price of new grocery PNP pallets.

PRM = price of raw material inputs used in production of new grocery

pallets.

PO = price of other factor inputs to the pallet production function. 12ll

inventory quantity QINV = of pallets available in the system.

Xt = dollar volume of grocery and related products sales in period t.

Using the variables defined above, the demand and supply functions

are summarized as followsz

Grocery and related products:

Retail Stores: QDGP Demand: = f { PGPS, PS, PC, PO, IC }

Grocery Distribution Centers: S

Demand. QD f { SC, GP,DC - PGPP, PGPS, HC, PP } Grocery Manufacturers: S ‘ S“PPlY· 6p,m -· { "¤pp· P¤M· pm- Pp- Po } ° Pallets: Demand(total). QD — P f { PP, PGPP, PGPS, Po } S Supply(total). Q = f { PP, PO } P PCAP, PLAB, Demand(new): QSP QDNP = - QINV (Excess Demand) QSNP Supply(new): = f { Po } PNP, PRM, Q Inventory: f {Xt (1 QINV = - - w )Xt_1} (Stock Adjustment) Market Clearing _ _ Identities. QD QS and D S P ; P Q NP ; Q NP CHAPTER VI

PALLET CONSUMPTION BY THE GROCERY INDUSTRY

The initial section of this chapter describes the available data

base and includes an analysis of the flows of grocery and related

products in terms of dollar volume of retail sales between market areas

within a region as well as market areas in different regions. The

purpose of this analysis is to provide a base of information from which

regional pallet consumption may be estimated under specific food flow

and pallet durability assumptions. Available price and quantity data

for grocery pallets is inadequate for statistical estimation of the

demand and supply models presented in the preceeding chapter.

Therefore, estimation of grocery pallet consumption, which does not consider pallet price in estimating quantity of pallets consumed, is presented in place of grocery pallet demand and supply estimation.

DATA BASE

The 1977 Census of Transportation contains the latest available data from the Bureau of the Census (U.S. Department of Commerce 1981) that identify the movement of goods between regions in terms of tons and ton-miles of shipments for three-digit SIC categories of food and kindred products. However, more recent and sufficiently detailed information published in the Progressive Grocer's 1987 Marketing

Guidebook (Progressive Grocers Information Sales 1987) will be used to

125 126

establish the levels of shipments within and between market areas. The

nine major regions identified by the Bureau of the Census in the United

States can be compared to the seven regions specified by Progressive

Grocer's Marketing Guidebook (Figure 6).

Although there are similarities between the regions identified by

the Bureau of the Census and the Marketing Guidebook, some major

differences exist. These differences are largely explained by the fact

that in the definition of regions, the Bureau of the Census uses state

boundaries to outline its regions while the Marketing Guidebook crosses

state boundaries to include adjoining counties in defining market

areas. First, the Bureau of the Census East South Central region and

portions of the South Atlantic regions are combined into the South East

region in the Marketing Guidebook. Second, the Bureau of the Census

Mountain region is partitioned between the West Central, South West, and

Pacific regions in the Marketing Guidebook. Third, the East Central and

Mid Atlantic regions in the Marketing Guidebook cover portions of the

Bureau of the Census South Atlantic region. Finally, the East Central region includes the Pittsburgh market area which is included in the

Bureau of the Census Mid Atlantic region.

Within each region of the country specified in the Marketing

Guidebook, major market areas are identified (Table 2). As an example, the East Central region is divided into eight market areas as follows:

Charleston/Roanoke, Cincinnati, Cleveland, Detroit, Grand Rapids,

Indianapolis, Louisville, and Pittsburgh. 127

Marketing Guidebook Regions: 0üENGLA I

Bureau of the Census Regions: — ENGLAND}usw

evI 1 F

ürFigure6. Comparison of Bureau of the Census and Marketing Guidebook Regions. 128

Table 2. Regions defined in the 1987 Marketing Guidebook.

Region: Market Areas:

New England Boston; Hartford

Mid-Atlantic Albany; Baltimore/Washington; Buffalo; New York; Philadelphia; Richmond

Southeast Atlanta; Birmingham; Charlotte; Columbia, SC; Jacksonville; Knoxville; Memphis; Miami; Nashville; New Orleans; Raleigh; Tampa

East Central Charleston/Roanoke; Cincinnati; Cleveland; Detroit; Grand Rapids; Indianapolis; Louisville; Pittsburgh

Southwest Albuquerque; Dallas; Houston; Oklahoma City; . San Antonio; Springfield

West Central Billings; Chicago; Denver; Des Moines; Fargo Kansas City; Milwaukee; Minneapolis; Omaha; St. Louis; Wichita

Pacific Alaska; Hawaii; Los Angeles; Phoenix; Portland; Sacramento/Fresno; Salt Lake City; San Francisco; Seattle; Spokane

Source: Progressive Grocers Information Sales. 1987. 1987 Marketing Guidebook. Stamford, Conn. 129

These market areas generally cover the standard metropolitan

statistical areas and surrounding counties. Using the

Charleston/Roanoke market area as an example, the market area is

comprised of N Kentucky counties, 3 Ohio counties, 30 West Virginia

counties, and 30 Virginia counties, surrounding the metropolitan areas

of Charleston, W. Va., and Roanoke, Va. Market information for each of

the counties includes statistics such as population, number of

households, dollar volume of food store sales, and number of food

stores, convenience stores, and supermarkets.

The 1986 dollar volume of food store sales for each market area

and the market area's percent share of the region's dollar volume of

food store sales are reported in the Marketing Guidebook (Table 3). The

individual market areas of New York and Los Angeles are the two largest

in terms of dollar volume of food store sales with $20.79 and $20.75

billion worth of sales, respectively. The dollar volume of food store

sales in each of these market areas exceeds that for the combined two market areas making up the New England region, which has the smallest level of sales at 15.ü9 billion dollars. The South East region, with 12 market areas, has the largest number of market areas of any region and the largest total dollar volume of food store sales at 63.79 billion dollars. Both the South East and the East Central regions are comprised of market areas with fairly uniform levels of sales, ranging generally between 3 and 8 billion dollars. The other five regions have at least one market area with sales exceeding 10 billion dollars and show 130

Table 3. Food Store Sales in 1986, by Region & Market Area.

PERCENT OF REGION MARKET AREA REGION SALES ($000) New England Boston 76.0 $11,788,944 Hartford 24.0 3,698,962 ToTAL 100.0 $15,487.906 Mid-Atlantic Albany 10.0 $5,607,955 Baltimore/Washington 20.0 11,170,217 Buffalo 7.4 4,101,796 New York 37.2 20,789,047 Ph11ade1ph1a · 17.0 9,468,848 R1chm¤¤a 8.4 4,685,126 TOTAL 100.0 $55.822,989 South East Atlanta 9.7 $6,184,960 s1r¤1¤gh¤m 8.4 5,358,737 Charlotte 5.4 3,460,527 Columbia, S.C. 10.2 6,499,413 Jacksonville 5.4 3,435,189 Knoxville 4.6 2,955,056 Memphis 11.5 7,308,076 M1¤m1 8.8 5,636,713 Nashville 4.8 3,095,671 New Orleans 10.2 6,480,781 Raleigh 8.1 5,181,349 Tampa 12.9 8,197,408 TOTAL 100.0 $63,793,880 East Central Charleston/Roanoke 9.5 $4,002,146 Cincinnati 16.7 7,046,424 Cleveland 15.1 6,328,266 Detroit 18.4 7.726,058 Grand Rapids 7.9 3.310.179 Indianapolis 10.7 4.512.758 Louisville 8.2 3,458,105 Pittsburgh 13.5 5,676,087 TOTAL 100.0 $ 2,060,023 131

Table 3. continued.

PERCENT OF REGION MARKET AREA REGION SALES ($000) West Central Billings 2.7 $1,¤1u,3l11 Chicago 21.6 11,282,#68 Denver 12.2 6,#01,808 Des Moines 6.0 3,120,##3 Fargo 3.0 1,592,231 Kansas City 7.9 #,108,853 Milwaukee 12.0 6,299,715 Minneapolis 9.7 5,0#2,066 Omaha 6.7 3,518,9#7 St. Louis 1#.9 7,767,359 Wichita 3.3 1,110,722 TOTAL 100.0 $52,258,953 South West Albuquerque 1#.0 $#,887,60# Dallas 29.6 10,3#3,536 Houston 21.0 7,318,515 Oklahoma City 13.# #,683,653 San Antonio 15.9 5,5#9,917 Springfield 6.1 2,132,660 TOTAL 100.0 $3 .915,885 Pacific Alaska 1.# $793,658 Hawaii 2.2 1,215,175 Los Angeles 36.8 20,750,916 Phoenix 8.1 #,558,119 Portland 7.5 #,20#,871 Sacramento/Fresno 10.8 6,087,5#8 Salt Lake City 7.3 #,123,0#1 san F1~anc1sc¤ 1#.5 8,199,19ls Seattle 7.1 #,013,756 Spokane #.3 2,#05,##0 ·r0·rAr. 100.0 $56,351,718

GRAND TOTAL $320,691,35#

Source: Progressive Grocers Information Sales. 1987. 1987 Marketing Guidebook. Stamford, Conn. 132

considerably more disparity between market areas with the largest and

smallest dollar volumes of sales.

Based on information published in the Marketing Guidebook, the

retail market shares for distribution centers serving a specific market

area can be identified. This information includes whether the

distribution center is located within the market area or is located

outside the market area. It also indicates the number of supermarkets

served in the area and in other market areas. For example, the

Charleston/Roanoke market area is served by a total of eighteen

distribution centers, of which eight are located within the market area

and ten are located outside the market area (Table N). These distribution centers serve a total of 191 chain supermarkets and 267

independent supermarkets with a total retail share of over 93 percent of

the total market, divided into 62.7 percent from in-market distribution centers and 31.1 percent from out—of-market distribution centers.

The out-of-market distribution centers serving a particular market area are not necessarily located in the same region as the market area they serve. In the case of the Charleston/Roanoke market area, the ten . out-of-market distribution centers are divlded as follows: two are located in the same region (East Central) as the Charleston/Roanoke market area with a combined retail market share of 3.91 percent; three are located in the Mid Atlantic region with a combined retail market share of 6.67 percent; and, the remaining five are located in the South

East region with a combined retail market share of 20.56 percent. ‘ 133

Table #. Market Area: Charleston/Roanoke

In-Market Distribution Centers:

Distribution centers Supermarkets Retail Supermarkets located within area in area Share Served

Acme Markets 8 2.12% 9 Associated Grocers, Inc. 23 #.17 23 Fleming Foods of VA 3# 5.82 52 The Kroger Co. 79 2#.73 107 Mid—Mountain Foods 19 6.11 #3 United Grocers, Inc. 29 5.12 29 Virginia Foods of Bluefield 19 2.82 19 Wetterau, Inc. 72 11.78 107

0ut—of-Market Distribution Centers: Supermarkets Retail Distribution centers located outside area in area Share

Atlantic & Pacific Tea Co. 6 1.02% Big Bear Stores 7 2.78 Fleming Foods of TN 28 #.02 Food Lion Stores 28 7.65 Harris-Teeter Spr. Mkts. 13 3.11 Merchants Distributors 11 2.38 Richfood, Inc. 25 #.65 Safeway Stores # 1.00 Super Valu Stores 9 1.13 Winn-Dixie Stores, Inc. 1# 3.#0

Source: Progressive Grocers Information Sales. 1987. 1987 Marketing Guidebook. Stamford, Conn. 13ü

The retail sales in 1986 attributed to distribution centers

serving each market area may be expressed as a percent of total food

store sales in the market area (Table 5) or as an actual dollar volume

of sales in the market area (Table 6). In Table 5 and Table 6, the

percent or dollar volume of total food store sales attributed to

distribution centers is separated into two catagories. These catagories

identify the location of distribution centers serving a particular

market area. The distribution centers serving a market area are either

located within the market area (IN—MARKET) or are located outside the

market area (OUT-OF-MARKET). The sum of these two catagories yields the

· total retail sales, expressed either as a percent or as a dollar volume,

attributed to all distribution centers serving a particular market area.

For example, Table 5 shows that in the Boston market area, 88.3 percent

of retail sales in 1986 can be attributed to distribution centers

located within the Boston market area. Another 8.2 percent of retail

sales can be attributed to distribution centers located outside the

Boston market area. The percent of total retail sales attributed to

distribution centers serving the Boston market area is the sum of the

in-market and out-of market percents, or 96.5 percent.

The combined retail market share for both in-market and

out-of-market distribution centers serving a particular market area does

not necessarily equal total food store sales in that market area. In

addition to those distribution centers serving the retail food stores in

an area, there are usually a number of in-market buying offices, food

( 135

Table 5. Retail Sales (X) Attributed to Distribution Centers, 1986.

DISTRIBUTION CENTER LOCATION IN- OUT-OF- REGION MARKET AREA TOTAL MARKET MARKET (X) (X) (X) New England Boston 96.50 88.30 8.20 Hartford 95.58 61.23 3ü.35

Mid-Atlantic — Albany _ 86.21 69.75 16.46 Baltimore/Washington 92.97 82.12 10.85 Buffalo 95.32 88.61 6.71 New York 95.N8 92.88 2.60 Ph11ada1ph1a 99.55 61.46 38.09 Richmond 95.19 79.01 16.18

South East Atlanta 93.15 46.26 46.89 Birmingham 96.5N 76.NO 20.1ü Charlotte 97.15 81.8N 15.31 c61um61a, s.c. 98.79 74.47 24.32 Jacksonville 92.75 67.17 25.58 Knoxville 91.06 49.71 41.35 Memphis 91.13 79.8ü 11.29 M1am1 96.5N 83.13 13.41 Nashville 91.96 76.62 15.3H New Orleans 96.76 69.55 27.21 Ra1a1gh 96.80 44.80 52.00 Tampa 93.65 77.08 16.57

East Central Charleston/Roanoke 93.81 62.67 31.1ü c1¤c1¤aa61 93.41 81.06 12.35 Cleveland 95.77 85.77 10.00 Detroit 90.79 60.25 30.5ü Grand Rapids 93.87 72.90 20.97 Indianapolis 95.96 69.83 26.13 Louisville 87.27 65.65 21.62 P16cabupgh 90.93 74.51 16.42 136

Table 5. continued.

DISTRIBUTION CENTER LOCATION IN- OUT-OF- REGION MARKET AREA TOTAL MARKET MARKET (X) (X) (X) West Central Billings 93.19 72.09 21.10 ch1cag¤ 92.79 84.65 8.09 Denver 97.22 99.86 2.36 Des Moines 96.92 82.95 13.97 Fargo 93.63 76.19 17.99 Kansas City 96.36 80.76 15.60 Milwaukee 99.65 87.70 6.95 Minneapolis 96.62 88.79 7.88 Omaha 95-79 57-22 38-57 sc. L6u1s 92.10 77.91 19.69 Wichita 95.68 72.35 23.33

South West Albuquerque 97.92 78.69 19.28 Dallas 92.38 84.63 7.75 Houston 89.75 83.82 5.93 Oklahoma c1uy 96.04 83.36 12.68 San Antonio 59.19 33.71 25.98 Springfield 99.71 76.69 18.07

Pacific Alaska 93.65 35.86 57.79 uawa11 73.99 27.88 45.61 Los Angeles 97.79 96.70 1.09 Phoenix 98.09 98.09 0.00 Portland 95.86 99.51 1.35 Sacramento/Fresno 80.45 32.53 97.92 Sale Lake City 97.05 91.03 6.02 San Francisco 85.01 72.73 12.28 Seattle 98.63 93.68 9.95 Spokane 95.29 51.86 93.93

Source: Progressive Grocers Information Sales. 1987. 1987 Marketing Guidebook. Stamford, Conn. 137

Table 6. Retail Sales ($000) Attributed to Distribution Centers, 1986.

DISTRIBUTION CENTER LOCATION IN- 0UT—0F- REGION MARKET AREA TOTAL MARKET MARKET ($000) ($000) ($000) New England Boston 11,376,331 10,ü09,638 966,693 Hartf0rd 3,535, 68 2,26 ,81ü 1,210,593 SUB-TOTAL 1 ,911,799 12,67 ,512 2,237,287 Mid-Atlantic Albany ü,83N,618 3,911,5ü9 923,069 Baltimore/Washington 10,38ü,951 9,172,982 1,211,969 Buffalo 3,939,832 3,63U,601 215,231 New York 19,8 9,382 19,308,867 5 0,515 Philadelphia 9,226,238 5,819,55U 3,606,68ü Richmond , 59,111 3,101,118 158,053 SUB-TOTAL 52,86 ,792 5,5 9,271 7,315,521 SQuth East Atlanta 5,761,290 2,861,162 2,900,128 Birmingham 5,173,325 u,09u,075 1,079,250 Charlotte 3,361,902 2,832,095 529,807 Columbia, S.C. 6,N20,770 N,8U0,113 1,580,657 Jacks¤¤v111e 3,186,138 2,307,416 878,721 Knoxville 2,690,87N 1,U68,958 1,221,916 M6mph1s 6,659,850 5,834,768 825,082 M1¤m1 5,Uü1,683 4,685,800 755,883 Nashville 2,8H6,779 2,371,903 N7ü,876 New Orleans 6,270,804 4,507,383 1.763,N21 Ra1e1gh 5,015,5ü5 2,321,244 2,694,301 Tampa 1,616,813 6,318,562 1,358,311 SUB-TOTAL 60,505,833 ,Uü3,N81 16,062,352 East Centra1 Charleston/Roanoke 3,75ü,ü13 2,508,1ü5 1,2ü6,268 c1nc1¤nac1 6,582,065 5,711,831 870,233 Cleveland 6,060,580 5,N27,75N 632,827 3222323 ,2 ran a S 2*32**322, 7, 3*333*338. , 2*233*%, 5 1¤d1a¤apg11s 4,330,443 3,151,259 1,179,184 L¤u1sv111e 3,017,888 ä,270,246 747,642 Pittsb gh ur 5,161,266 ,229,252 932,013 suB—T0TAL 39,028, 08 30,366,558 8,661,850 138

Table 6. continued.

DISTRIBUTION CENTER LOCATION IN- OUT-OF- REGION MARKET AREA TOTAL MARKET MARKET ($000) ($000) ($000) West Central Billings 1,318,029 1,019,598 298,926 cmcago 10 , 963, 361 9 ,550, 609 912,752 Denver 6,223,838 6,072,755 151,083 Des Moines, 3,029,333 2,588,907 935,926 Fargo ~ 1,990,806 1,212,325 278,981 Kansas City 3,959,291 3,318,310 690,981 Milwaukee 5,962,680 5,529,850 937,830 N1¤neap¤11S 9,871 , 61111 9,979,329 397,315 Omaha 3.370.799 2.013.591 1.357.258 St. Louis 7,153,138 6,012,713 1,191,025 wicmca 1,636, 19 1,231,101 399,111 SUB-TOTAL 9. 75.333 3.025. 195 6. 50. 188 ' South West Albuquerque 9,785,992 3,893,612 992,330 Dallas 9 . 555 . 359 8 . 753 . 735 801 . 629 Houston 6,568,367 6,139,379 933,988 Oklahoma City 9,998,180 3,909,293 593,887 San Antonio 3,289,936 1,873,377 1,919,119 Springfield 2,019, 2 1,63 , 11 385,312 SUB-TOTAL 30,712,686 26,191,366 ,571,320 Pacific Alaska 793,261 289,606 958,655 Hawaii 893,032 338,791 559,291 Los Angeles 20,292,321 20,066,136 226,185 Phoenix 9,968 , 780 9,968 , 780 0 Portland 9,030,789 3,979,029 56,766 Sacramento/Fresno 9,897,932 1,980,279 2,917,153 Salt Lake City 9,001,911 3,753,209 298,207 San Francisco 6,970,135 5,963,279 1,006,861

Spokane 2,292,1 1,2 1, 1 1,0 , 3 SUB-TOTAL 52,598,073 95,836,6 1 6,711, 32 GRAND TOTAL 300,0116,9211 2118,036,9111 52,009,950

Source: Progressive Grocers Information Sales. 1987. 1987 Marketing Guidebook. Stamford, Conn. 139

brokers, non-food distributors, candy, tobacco and media distributors

who provide the balance of products to the food stores. What is

important is that the combined retail market share for the distribution

centers averages over 93 percent of total retail food store sales for

all market areas in the United States. This means that the vast

majority of grocery and related products delivered to retail food stores

are handled in a distribution system which has been shown to use pallets

in accomplishing those deliveries.

Examination of retail sales expressed as a percent of total sales

shows that the average share of retail sales over all market areas is

73.33 percent for the in-market distribution centers and is 19.85

percent for the out-of-market distribution centers. Sales volume

attributed to the in—market distribution centers ranges from a low of

27.88 percent for the Hawaii market area to a high of 98.0ü percent for

the Phoenix market area. The sales volume attributed to out—of—market

distribution centers ranges from a low of 0.0 percent for the Phoenix

market area to a high of 57.79 percent for the Alaska market area. - The Alaska and Hawaii market areas are unique in that they are

physically separated from the other market areas in the continental

United States. Retail stores in both market areas receive more

shipments from distribution centers located in market areas along the

West Coast than they do from distribution centers located in their

respective market areas. This may be explained by the economies of

scale achieved by the distribution centers on the West Coast, serving ” 1NO

much larger markets in terms of sales volume than either the Alaska or

Hawaii market area.

The Phoenix market area is something of an anomaly in that it is

served by no out—of-market distribution centers. In this case, the

market is large enough for in—market distribution centers to operate

efficiently and is concentrated largely in the Phoenix/Tucson area,

which means that there are no other distribution centers within cost

effective shipping distance to this market.

Market concentration and economies of scale in distribution

systems do have an effect on the amount of retail sales which can be

attributed to out-of—market distribution centers. In the two largest

markets, New York and Los Angeles, the percent of retail sales allocated

to out-of-market distribution centers is 2.60 percent and 1.09 percent,

respectively. In the other four market areas with retail sales over 10

billion dollars, the percent of retail sales allocated to out-of-market

distribution centers ranges from 7.75 percent to 10.85 percent, which is

well below the average for market areas over the entire United States.

ESTIMATION OF PALLET CONSUMPTION

Based on the stock adjustment model for grocery distribution, new

pallet censumption is expressed in terms of the inventory quantity of

pallets in the system, or ’ G °N1>,c QINV,t ' (1'" )QINV,t-1 IH1

where is the quantity of QNP’t new pallets consumed in materials

handling of grocery and related products in period t, is the QINv't quantity of pallets in inventory in period t, and (1-w*) is the

percent of pallets remaining in inventory from the preceeding period.

Estimation of the quantity of new pallets consumed in period t therefore

requires identification of the quantity of pallets in inventory in

successive time periods. Total pallet production data and total retail

grocery store sales are available over the past 13 years (Table 7).

However, this information has certain limitations that preclude

statistical estimation of the inventory quantity of pallets in the

system. Total pallet production data include all types of pallets

produced, that is, expendable as well as permanent pallets of varying

size and construction specifications. Thus, total pallet production

includes many factors, only one of which relates to the consumption of

pallets by the grocery industry. For example, the increase in total

pallet production from 228.3 million pallets in 1982 to 372.8 million pallets in 1986 represents an increase of over 63 percent in production over the four year period. This increase can be attributed largely to growth in the general economy. As industrial production has expanded during this period, the demand for pallets by all industries using pallets in materials handling has increased as well. Thus, the 27.3 percent increase in grocery store sales over the same period, measured in current dollars, can be expected to account for part of the increase in total pallet production; but, the exact amount of increased pallet 1l|2

Table 7. Pallet Production and Grocery Store Sales.

Year Pallet Production Grocery Store Sales for all uses Total Units Total

(million) (current $ billion)

197M 205.1 130.5 1975 159 3 192 5 - - 1976 195.7 152.0 1977 235.9 162.8 1978 270. 3 179.6 1979 296- 0 199 - 9 1980 258.ü 220.8 1981 252.2 2M0.9 1982 228.3 252.0 1983 257.7 263.8 1981+ 301+ . 1+ 279 .1+ 1985 33ü.8 292.2 1986 372.8 320.7

Sources: National Wooden Pallet and Container Association. 1986. Production Report (New Pallets). Washington, D.C., and Progressive Grocers Information Sales. 1987. 1987 Marketing Guidebook. Stamford, Conn. „ 1N3

production attributed to increased grocery store sales cannot be

identified.

In estimating quantity of new pallets consumed in the grocery

distribution system, the portion of total pallet production that is used

to maintain the inventory quantity of pallets in the grocery and related

products manufacturing and distribution industry must be identified.

Point estimates of total ü8xü0 pallet production, which include grocery

Vpallets but are not limited to grocery pallet production, are available

from surveys taken in three specific years, 1979, 1982, and 198M. These

estimates range from a low of 20 percent of total pallet production in

1982 (Mckeever, et al. 1986) to a high of 31 percent in 198ß (NWPCA

1985) with an estimate of 27 percent in 1979 (NWPCA 1980). These

estimates are incomplete measures of grocery pallet production because

the U8xN0 size pallet is used in materials handling of products other

than grocery and related products. Although grocery pallet production

estimates are contained in the point estimates, the actual quantity of grocery pallets produced is not reported.

Because neither the quantity of new grocery pallets nor the total inventory quantity of pallets used by the grocery distribution system can be identified directly from the available data, estimation of the inventory quantity of pallets in the system must be made indirectly through assumptions about the relation of retail sales to inventory quantity of pallets. Estimation is based on assumptions about the nature of pallet use in the system from a study by Wallin (1977) and wh

information obtained during interviews with grocery distribution center

managers.

CONVERSION FACTORS

The quantity of grocery and related products demanded in a

particular area is satisfied by shipments of products from distribution

centers to retail stores. The number of shipments from distribution

centers is therefore directly reflected in the dollar volume of retail

sales in an area. This measure of shipments, that is, retail sales, can

be converted into an estimate of the number of pallets required to

sustain that level of shipments. Certain assumptions about pallet use

in the grocery distribution system must be made to develop the

conversion factors and make estimates of the number of pallets in the

system. These assumptions include the following:

a. Average pallet life = 2.5 years.

b. Average trips per life of pallet = 20 trips.

c. Average number of pallets per shipment = 22 pallets.

d. Average retail dollar volume per shipment = $15,000.

e. Average turnover period for product in warehouse = 3 weeks.

The first two assumptions are based on data from an unpublished report by Wallin (1977). The last three assumptions are based on information obtained from the survey of grocery distribution center managers. The assumption about dollar volume per shipment to retail stores is critical 145

because it provides the link between known levels of retail sales and

the number of pallets in use in the system.

From Table 7, one notes that 320.7 billion dollars in retail

grocery sales were reported in the United States in 1986. But, only

93.6 percent of these sales were attributed to retail stores served by

distribution centers, or just over 300 billion dollars based on

information from the 1987 Marketing Guidebook. Assuming that these

sales required the use of trucks to deliver the products from

distribution centers to retail stores, then an estimated 20 million

truck loads were required to deliver the grocery products to retail

stores, at an average of $15,000 of product per truck load. With an

average of 22 pallets per truck load, an estimated total of NÄO million

pallet loads were required to move the products. Based on the

assumptions that a pallet remains in the system for 2.5 years, on the

average, and that it makes an average of 20 trips in that time, it is

assumed that the average pallet makes 20/2.5 or 8 trips per year.

Therefore, the HNO million pallet loads required an estimated 55 million pallets, with each pallet making a minimum of 8 trips, to deliver grocery and related products to retail stores in 1986.

One may approach this estimate of 55 million pallets in another way. Industry sources indicate that in 1986 the dollar value per pallet load of product shipped from a distribution center varied from approximately $2000 per pallet for meat to $350 per pallet for produce, with an average value of products carried on each pallet equal to 1ü6

approximately $700.Q Using the above assumptions, it is calculated that

each pallet carries an average of $15,000/22 or $682 of grocery products

on each trip, which is reasonably close to the average value per pallet

load reported by industry sources. Again assuming that each pallet

makes 8 trips per year, the total retail sales dollar volume for each

pallet used in shipping grocery products to retail stores is 8 times

$682, or $5N56 per pallet. Thus, in satisfying 300 billion dollars in

retail sales, distribution centers needed 300 billion/SN56, or about 55

million pallets to move grocery and related products from the

distribution centers to retail stores.

The above estimate omits two important aspects of grocery

distribution. First, the pallets used in movement of products from the

manufacturer to the distribution center must be included, and second,

the number of pallets used in the warehouse for product storage must be

estimated.

Further information is required to estimate the number of pallets

used in movement of products from the manufacturer to the distribution

center. For example, if it is assumed that the same number of pallets

are required to deliver the products to the distribution centers as are

required to deliver products to the retail stores, then another 55

million pallets could be added to the number of pallets required for the movement of grocery and related products. However, based on the survey of grocery distribution center managers, about 35 percent of the products coming in to the distribution center are not palletized. Thus, 1117

in 1986 the number of pallets required for movement of grocery and

related products from the manufacturers to the distribution center is

estimated to be 65 percent of the number required for movement of

products to the retail stores, or about 36 million pallets.

Estimation of the number of pallets used in the warehouse for

product storage is based on the assumption of the frequency of product

turnover in the warehouse and how it relates to the number of pallets

used to distribute products to the retail stores. If it is assumed that

the product turnover occurs every three weeks, or on an average of 17

times per year as reported by the surveyed distribution center managers,

then the dollar volume of products stored in the warehouse is assumed to

be 3/52 or about 6 percent of the total dollar volume of products moved

from the distribution center to the retail stores in a period of one

year. Six percent of the 55 million pallets used to move products to

the retail stores yields an estimated 3 million pallets in the

distribution system for the storage of products in the warehouse. This

estimate is reinforced by the observed quantity of pallets in surveyed

distribution centers. For example, one large distribution center

shipped approximately 1.7 million pallet loads of products to retail stores in 1986. This distribution center maintained an average of

100,000 pallets in the warehouse for storage of grocery and related products, which is about 6 percent of the quantity of pallets the distribution center used to move products to retail stores. 1118

The total inventory quantity of pallets in the grocery

distribution system is attained by adding up the estimates made from the

above assumptions. Again, using 1986 retail sales to base the

assumptions on, the total number of pallets in the system is estimated

to be 55 + 36 + 3 million pallets or 9N million pallets. This is the

total inventory quantity of pallets in the grocery distribution system

required for the movement and storage of grocery and related products in

1986, not the quantity of new pallets required in 1986.

In order to obtain an estimate of the quantity of new pallets required in 1986, one must first estimate the inventory quantity of pallets in the system in the preceeding year, 1985. With the same assumptions outlined above, the total inventory quantity of pallets required in 1985 to handle the volume of products generating 273 billion dollars in retail sales is calculated to be 85 million pallets. Thus, if all pallets in the system in 1985 remained in the system in 1986, then an additional 9 million new pallets were required to move the increased volume of grocery products. However, based on the assumption that a pallet stays in the system for 2.5 years, an estimated UO percent of the pallets in the system were discarded in 1985. Therefore, an additional 3ü million new pallets were required to replace pallets discarded in 1985. Using the equation specified earlier, I calculate

= 9N (0.60)*85 M3 QNP’1986 - = 199

Total new grocery pallets purchased or consumed in 1986 is therefore

estimated to be N3 million pallets, or over 11 percent of total pallet

production in 1986.

The only published reference points for validating this estimate

of grocery pallet production are the point estimates for ü8xü0 pallet

production reported above. Because these reported percentages include

all uses of N8xNO pallets, the estimate of grocery pallet production was

expected to be less than the reported percentages. Dr. Edward C.

Brindley, Jr., President, Industrial Reporting, Inc., Richmond, Va.,

estimated that grocery pallet production in Virginia was less than 15

percent of total quantity of pallets produced in the state. Brindley

also estimated that national levels of grocery pallet production are

about half of the total market for the ü8xü0 pallet. Based on the

reported surveys, this would indicate that grocery pallet production

could range from 10 to 15 percent of total production. Thus, the

estimate of #3 million grocery pallets purchased in 1986 appears

reasonable in view of the existing information.

REGIONAL PALLET CONSUMPTION

The volume of retail sales of grocery and related products is the driving force in generating consumption of new pallets. Without retail sales there would be no incentive to move or store grocery and related products on pallets. For this reason, the estimate of consumption of new grocery pallets in 1986 is broken down between regions on the basis 150

of the volume of retail sales occurring in each region. From the

previous section, I note that 38 percent of new pallet consumption

occurs in the movement of grocery products to distribution centers,

while 62 percent occurs in the storage and movement of grocery products

through distribution centers. As a result, allocation of the quantity

of new pallets consumed between regions depending on regional retail

sales is biased in that not all new pallets are consumed at distribution

centers in satisfying retail sales. In fact, 38 percent of new pallet

consumption occurs at grocery and related products manufacturing

facilities, not at distribution centers. Inasmuch as the grocery manufacturing facilities are not always located in the same region as

the distribution centers they serve, the regional estimates of new pallet consumption are not exact. However, these estimates do reflect the concentration of distribution centers within market areas and for that reason, they are included in this analysis.

In order to allocate the estimate of N3 million grocery pallets between regions, or more specifically, between market areas as defined earlier, regional levels of retail sales attributed to distribution centers must first be expressed as a percent of total sales (Table 8).

The percent of national retail sales attributed to distribution centers is separated into two catagories that identify the location of distribution centers serving a particular market area. For example, in the Boston market area, 3.ü7 percent of retail sales is attributed to 151

Table 8. Regional Sales, as a percent of Total Sales in 1986.

DISTRIBUTION CENTER LOCATION IN- OUT-OF- REGION MARKET AREA TOTAL MARKET MARKET New England (percent) (percent)- (percent) Boston 3.79 3.#7 0.32 Hartford 1.18 0.75 0.#3 SUB—TOTAL #.97 #.22 0.75 Mid—Atlantic Albany 1.61 1.30 0.31 Baltimore/Washington 3.#6 3.06 0.#O Buffalo 1.30 1.21 0.09 New York 6.62 6.## 0.18 Philadelphia 3.1# 1.9# 1.20 Richmond l.#Q 1.23 0.26 SUB-TOTAL 17.62 15.18 2.## South East Atlanta 1.92 0.95 0.97 Birmingham 1.72 1.36 0. 36 Charlotte 1.12 0.9# 0.18 Columbia, S.C. 2.1# 1.61 0.53 Jacksonville 1.06 0.77 0.29 xaaxv111a 0.90 0.49 0.41 Memphis 2.21 1.9# 0.27 M1am1 1.81 1.56 0.25 Nashville 0.95 0.79 0.16 New Orleans 2.09 1.50 0.59 Raleigh 1.67 0.77 0.90 Tampa 2.56 2.11 0.45 SUB—TOTAL 20.15 1#.79 5.36 East Central Charleston/Roanoke 1.26 0.84 0.#2 Cincinnati 2.19 1.90 0.29 Cleveland 2.02 1.81 0.21 Detroit 2 . 3# 1 . 55 0 . 79 Grand Rapids 1.03 0.80 0.23 Indianapolis 1.## 1.05 0.39 Louisville 1.01 0.76 0.25 Pittsburgh 1.72 1.#1 0.31 SUB-TOTAL 13.01 10.12 2.89 152

Table 8. continued.

DISTRIBUTION CENTER LOCATION IN- OUT-OF- REGION MARKET AREA TOTAL MARKET MARKET West Central (percent) (percent) (percent) Billings 0. ua 0.3u 0.10 Chicago 3.U9 3.18 0.31 Denver Des Moines 2.071.01 0.262.02 0.150.05 Fargo 0.50 0. 1 0.09 Kansas City 1.32 1.11 0.21 Milwaukee 1.99 1.8N 0.15 Minneapolis 1.62 1.ü9 0.13 Omaha 1.12 0.67 0.ü5 St. Louis 2.38 2.00 0.38 Wichita 0.E5 0.ü2 0.13 SUB-TOTAL 16. 9 1Ü.3N 2.15 South West Albuquerque 1.60 1.28 0.32 Dallas 3.19 2.92 0.27 Houston 2.19 2.05 0.1N Oklahoma City 1.50 1.30 0.20 San Antonio 1.09 0.6ä 0.N7 Springfield 0.6% 0.5 0.13 SUB-TOTAL 10.2 8.71 1.53 Pacific Alaska 0.2M 0.09 0.15 Hawaii 0.30 0.12 0.18 Los Angeles 6.77 6.69 0.08 Phoenix 1.ü9 1.U9 0.00 Portland 1.3N 1.32 0.02 Sacramento/Fresno 1.63 0.66 0.97 Salt Lake City 1.33 1.25 0.08 San Francisco 2.33 1.99 0.3N Seattle 1.32 1.ä5 0.07 Spokane 0.]] O. 2 0.35 _ SUB-TOTAL 17.52 15.28 2.2ü

GRAND TOTAL 100.00 82 . 614 1] .36

Source: Progressive Grocers Information Sales. 1987. 1987 Marketing Guidebook. Stamford, Conn. 153 distribution centers located within the market area. Another 0.32 percent of retail sales is attributed to distribution centers located outside the Boston market area. The total percent of national retail sales attributed to distribution centers serving the Boston market area is the sum of the above values, or 3.79 percent of all retail sales attributed to distribution centers throughout the United States in

1986. As shown in Table 8, nearly 83 percent of retail sales attributed to distribution centers throughout the United States comes from distribution centers located within the particular market area served.

Multiplying the percent of retail sales in each market area by the total quantity of new pallets consumed yields the estimated quentity of grocery pallets purchased in each market area in 1986 (Table 9). The estimated quantity of new grocery pallets consumed is also separated into two categories that identify the location of distribution centers serving a particular market area. For example, in the Boston market area, an estimated 1.U9 million grocery pallets were consumed in 1986 by distribution centers located within the Boston market area. Another

0.1M million grocery pallets were consumed by distribution centers serving the Boston market area from outside the market area. Thus, total grocery pallet consumption estimated for distribution centers serving the Boston market area in 1986 was 1.63 million pallets. As shown in Table 9, over 35 million grocery pallets were estimated to be marketconsumedareain served.1986 by distribution centers located within the particular 150

Table 9. Estimated Regional Pallet Consumption in 1986.

DISTRIBUTION CENTER LOCATION IN- OUT-OF- REGION MARKET AREA TOTAL MARKET MARKET (million) (million) (million) New England Boston 1.63 1.09 0.1u Hartford 0.51 0.}} 0.18 SUB-TOTAL 2.1 1.82 0.32 Mid-Atlantic Albany 0.69 0.56 0.13 Baltimore/Washington 1.09“’ 1.32 0.17 Buffalo 0.56 0.52 0.00 New York 2.85 2.77 0.08 Philadelphia 1.35“’ 0.83 0.52 Richmond 0.60-/ 0.5} 0.11 SUB-TOTAL 7.58 6.53 1.05 South East Atlanta 0.83 0.01 0.02 V Birmingham 0.74 0.59 0.15 Charlotte 0.18/’ 0.01 0.07 Columbia, S.C. 0.92" 0.69 0.23 Jacksonville 0.06 0.33 0.13 7 “ Knoxville 0.39 0.21 0.18 Memphis 0.90·’ 0.83 0.11 M1a¤1 · 0.78 “ 0.67 0.11 N¤shv111e 0.111 0. 311 0.07 New Orleans 0.90 0.65 0.25 Raleigh 0.72 V 0.33 0.39 Tampa 1.10 0.91 0.19 SUB-TOTAL 8.67 6.37 2.30 East Central Charleston/Roanoke 0.50 V 0.36 0.18 c1¤c1¤nau1 0. 911 0. 82 0.12 Cleveland 0.87 0.78 0.09 Detroit 1.01 0.67 0.30 emma Rapids 0.1111 0.30 0.10 Indianapolis 0.62 0.05 0.17 Louisville 0.03 0.32 0.11 Pittsburgh O.]0V’ 0.61 0.1} SUB-TOTAL 5.59 0.35 1.20

ra J r'

155

Table 9. continued.

DISTRIBUTION CENTER LOCATION IN- OUT-OF- REGION MARKET AREA TOTAL MARKET MARKET (million) (million) (million) West Central Billings 0.19 0.15 0.0N Chicago 1.50 1.37 0.13 Denver 0.89 0.87 0.02 Des Moines 0.ü3 0.37 0.06 Fargo 0.21 0.17 0.0M Kansas City 0.57 0.H8 0.09 Milwaukee 0.85 0.79 0.06 Minneapolis 0.70 0.6ü 0.06 Omaha 0.+8 0.29 0.19 St. Louis 1.03 0.86 0.17 Wichita 0.2N 0.18 0.06 SUB-TOTAL 7.09 6.17 0.92 South West Albuquerque 0.69 0.55 0.1ü Dallas 1.37 1.26 0.11 Houston 0.9ü 0.88 0.06 Oklahoma City 0.6N 0.55 0.09 San Antonio O.ü7 0.27 0.20 Springfield 0.29 0.23 0.06 SUB-TOTAL . 0 3.7*+ 0.66 Pacific Alaska 0.10 0.0ü 0.06 Hawaii 0.13 0.05 0.08 Los Angeles 2.91 2.88 0.03 Phoenix 0.6ü 0.6ü 0.00 Portland 0.58 0.57 0.01 Sacramento/Fresno 0.70 0.28 0.ü2 Salt Lake City 0.57 0.5ü 0.03 San Francisco 1.00 0.85 0.15 Seattle 0.57 0.5N 0.03 Spokane 0.33 0.18 0.15 SUB-TOTAL 7 . 53 6. 57 0.96 GRAND TOTAL *+3.00 35.55 7.*+5 ' CHAPTER VII

EFFECT OF GROCERY PALLET USE ON HARDWOOD RESOURCE

This chapter describes the major characteristics of wood raw

material usage by the grocery distribution industry. The emphasis is on

wood raw material consumption in the construction of new grocery

pallets, although the quantity of wood raw material consumed in pallet

repair is also considered. Since hardwood pallets are presently

preferred for distribution of grocery and related products, the impact

of increasing quantity of grocery pallets on the hardwood resource is

discussed in a concluding section. Because of data limitations,

particularly with respect to overall pallet production within regions,

this section will be largely descriptive in nature.

In the grocery pallet supply model, the quantity of new pallets

supplied is expressed as a function of the price of pallets, price of

wood raw material input, and price of other factor inputs to the pallet

production function. Estimation of this model would show that price of

pallets and price of wood raw material directly effect the quantity of

pallets supplied. However, in this study, estimation of quantity of new

pallets supplied in grocery distribution is limited to a technical

conversion for quantity of new pallets consumed. This does not mean

that the interaction between prices of wood raw material and pallets and

the quantity of pallets supplied is not functioning in the system;

rather, the inadequacy of available data to estimate these interactions

156

I 157

necessitates a technical conversion approach to the estimation of wood

raw material consumed.

WOOD RAW MATERIAL - NEW PALLETS

The volume of wood raw material consumed annually in the

production of grocery pallets can be estimated through an analysis of

national and regional levels of new pallet consumption in grocery

distribution. In this analysis, the quantity of new pallets consumed is

converted to an equivalent volume of wood raw material input to production of grocery pallets. In making the conversion, both the volume of wood contained in the average grocery pallet and the volume of wood consumed in the production of the pallet are evaluated. I consider the board foot volume in an average grocery pallet nationally to be in the same range as that reported for non-expendable pallets by McCurdy

(1988), or 16.2N board foot per pallet. This value is the average volume of lumber contained in a non-expendable pallet, not the volume of wood raw material consumed in the production of the pallet.

Although McCurdy (1988) also reports that the average volume of lumber per pallet varies depending on the region where the pallet is produced, his study reflects all types of non-expendable pallets.

Unlike this study, it is not restricted specifically to grocery pallets. I consider that the modified GMA grocery exchange pallet

(described in Appendix C) constitutes the average pallet in the industry 158

and is used as a reference point by pallet manufacturers and purchasers

of grocery pallets throughout all regions of the United States.

In estimating the volume of wood raw material consumed in the

production of grocery pallets, I must consider the volume of wood lost

during production as well as the volume of wood actually contained in

the pallet. The wood volume lost during production includes sawdust,

edgings, and end trims, which are not included in the volume of wood

contained in the finished pallet. This wood volume lost during

production is expressed as a percent of the volume of wood actually

contained in the finished pallet. Pallet manufacturers contacted during

this study and in my previous studies relating to the pallet industry

indicate that the average waste factor for pallet production, or the

wood volume lost during production, is about 20 percent of the wood

volume actually going into pallets. This factor will vary from one

pallet manufacturer to another depending on the conversion efficiency of

each mill and the type of wood raw material used in the production

process, ie., logs, lumber, cants, or pre—cut parts. But, an industry

average of 20 percent waste in production appears realistic based on the

information available. Thus, a pallet containing 16.2M board foot of

wood requires an estimated 19.5 board foot of wood raw material in production of the pallet.

Although grocery pallets only constituted an estimated 11 percent of the total quantity of pallets produced in 1986, the volume of wood consumed in production of these pallets is shown to be a significantly ‘ 159 ~

greater share of total volume of hardwood raw material consumed in

pallet production. Based on the above estimate of wood consumed

perpallet,the M3 million pallets required by the grocery distribution

industry in 1986 consumed a total of 838.5 million board foot of wood

_ raw material. This is 18 percent of an estimated N.5 billion board foot

of hardwood raw material consumed by the industry for all types of

pallets in 1986 (McCurdy 1988).

Regional allocation of the volume of wood raw material consumed in

the production of grocery pallets is based on the estimated regional fä ä /s S consumption of new grocery pallets shown previously in Table 9.’/Äsß

noted earlier, these estimates cannot be considered exact because of the

uncertainty associated with the location of grocery manufacturing

facilities that use pallets in the movement of grocery products from the

manufacturing facility to distribution centers. However, the majority

of new pallets in grocery distribution are consumed at distribution

centers. This consumption has been shown to be satisfied by the supply

of pallets from pallet manufacturers in close proximity to the point of

demand (McCurdy and Ewers 1986). Therefore, in the absence of better

· information, the estimated regional new grocery pallet consumption is

used to provide a qualified estimate of the regional volume of wood raw

material consumed in the production of grocery pallets in 1986 (Table

10). The estimated quantity of wood raw material consumed is sepereted

into two categories that identify the location of distribution centers

serving a particular market area. For example, in the Boston market 160

Table 10. Estimated Wood Raw Material Consumption in Grocery Pallet Production, 1986.

DISTRIBUTION CENTER LOCATION IN- OUT-OF- REGION MARKET AREA TOTAL MARKET MARKET (MBF) (MMBF) (MMBF) New England Boston 31.g9 29.09 2.70 Hartford 9. 8 6.33 3.55 SUB-TOTAL 1.67 35. 2 6.25 Mid-Atlantic Albany 13.51 10.93 2.58 Baltimore/Washington 29.02/’ 25.63 3.39 Buffalo 10.93 10.16 0.77 New York 55.#7 53.96 1.51 Pn11ada1ph1a 16.26 10.08 R1anaand 12.#6„/ 10.34 2.12 SUB-TOTAL 1#7.73 127.29 20.## South East Atlanta 16.10 8.00 8.10 B1rn1ngnan 14.46 11.44 3.02 Charlotte 9.#0/ 7.91 1.48 Columbia, S.C. 17.94/ 13.53 #.42 Jacksonville 8.90 6.45 2.#6 xnaxv111a 7.52* #.11 3.41 Memphis 18.61/’ 16.31 2.31 Miami 15.21 13.09 2.11 Nashville 7•96“ 6.63 1.33 New Orleans 17.52 12.60 #.93 Ra1a1gh 1#.32~· 6.49 7.53 Tampa 21. 5 1%.66 3.80 SUB-TOTAL 169.09 12 .20 4.89 East Central Charleston/Roanoke 10.#9»/ 7.01 3.#8 Cincinnati 18.39 15.96 2.#3 Cleveland 16.9# 15.17 1.77 Detroit 19.60 13.01 6.59 Grand Rap1da 8.68 6.74 1.94 Indianapolis 12.10 8.81 3.30 Louisville 8.#3 6.3# 2.09 p1ttabnrgn 1#.#2/” 11.82 2.60 SUB-TOTAL 109.07 Q 84.86 24.21

\‘Q _; .1 ~ ' ~ 161

Table 10. continued.

DISTRIBUTION CENTER LOCATION IN- OUT—OF- REGION MARKET AREA TOTAL MARKET MARKET (MMBF) (MMBF) (MMBF) West Central Billings 3.68 2.85 0.83 Chicago 29.2# 26.69 2.55 Denver 17.39 16.97 0.#2 Des Moines 8.#5 7.23 1.22 Fargo #.17 3.39 0.78 Kansas City 11.06 9.27 1.79 M11wa¤kee 16. 66 15. Mu 1.22 Minneapolis 13.61 12.50 1.11 Omaha 9.#2 5.63 3.79 St. Louis 19.99 16.20 3.19 Wichita .51 3. 6 1.12 SUB-TOTAL 138.26 120.2 18.03 South West Albuquerque 13.37 10.7# 2.63 Dallas 26.70 2#.#6 2.2# Houston 18.36 17.1# 1.21 Oklahoma City 12.57 10.91 1.66 San Antonio 9.18 5.23 3.95 Springfield 5.6# #.5] 1.08 sur;-Tom;. 85.83 73.05 12.77 Pacific Alaska 2.08 0.80 1.28 Hawaii 2.50 0.95 1.55 Los Angeles 56.71 56.08 0.63 Phoenix 12.u9 12.49 0.00 Portland 11.26 11.11 0.16 Sacramento/Fresno 13.69 5.53 8.15 Salt Lake City 11.18 10.#9 0.69 San Francisco 19.#8 16.66 2.81 Seattle 11.06 10.51 0.56 Spokane 6.#1 3.#9 2.92 sua—·r0·rA1. 1E6.85 128.09 18.76

GRAND TOTAL 838.50 693.15 1#5.35 162

area, an estimated total of 31.79 million board foot of wood raw

material was consumed in 1986 in grocery pallet production for

distribution centers serving that market area. This total is further

broken down into 29.09 million board foot of wood raw material consumed

for distribution centers within the Boston market area and 2.70 million board foot of wood raw material consumed for distribution centers serving the Boston market area that are located outside the market area. As shown in Table 10, over 693 million board foot of wood raw material are estimated to be consumed in production of grocery pallets

for distribution centers located within the market area served by those distribution centers.

— WOOD RAW MATERIAL PALLET REPAIR

The volume of wood raw material consumed annually in the repair of grocery pallets depends on two technical factors and one economic factor. The two technical factors include the quantity of pallets actually repaired and the severity of damage sustained by pallets requiring repair. The economic factor is simply the interaction between the cost of pallet repair and the purchase price of a new grocery pallet. As the cost of repair increases relative to the purchase price of a new pallet, the grocery distribution system can be expected to replace more damaged pallets with new pallets rather than repair.

Conversely, as the purchase price of new grocery pallets increases relative to the cost of pallet repair, the grocery distribution system 163

can be expected to increase the extent of repair for pallets which would

earlier have been replaced with new pallets. I note that one of the

elements that is considered in both the cost of repair and the price of

new pallets is the price of wood raw material used in repair of damaged

pallets or in construction of new pallets. Thus, the price of wood raw

material is implicitly contained in the interaction between the cost of

pallet repair and the purchase price of new pallets.

Given a level of pallet consumption in grocery distribution as

noted in Chapter 6, I can estimate the volume of wood raw material

consumed based on an analysis of the above technical factors, i.e., the

quantity of pallets repaired and the severity of damage that requires

repair. As noted in Chapter 5, pallets are repaired on a more or less

continuous basis each year in order to maintain the inventory quantity

of pallets in the system. While precise figures are not available with

regard to the number of pallets repaired each year, estimates can be made on the basis of information provided by distribution center managers and our previous estimates of the inventory quantity of pallets in the system. Information provided by distribution center managers indicates that a typical distribution center either repairs in·house or contracts for outside repair of an average of 100 pallets per day.

Although large distribution centers (with over one million square feet of floor space) may repair 300 to 750 pallets daily, an average sized distribution center (with one-half million square feet of floor space) usually reported around 100 pallets repaired daily. Assuming that 100 16ü

pallets repaired daily represents an industry average, then

approximately 26,000 pallets are repaired each year in an average

distribution center. This calculation is based on the fact that a

typical pallet repair operation is reported to operate on a #0-hour,

5-day work week, or about 260 days per year. With NO9 distribution

centers in the system, the total quantity of pallets repaired in

distribution centers in 1986 is estimated to equal 10.6 million pallets.

However, pallets used for storage and movement of grocery products

through distribution centers are estimated in an earlier chapter to

constitute only 62 percent, or 58 million out of the total 9N million

pallets estimated in the grocery distribution system pallet inventory in

1986. The above 10.6 million pallets reflects a repair rate in

distribution centers of 18.28 percent of inventory quantity. This rate

of repair is considered to be the same for grocery manufacturers in maintaining the balance of 36 million pallets in the grocery distribution inventory in 1986. Recalling that these pallets were involved in moving grocery and related products from manufacturers to distribution centers, an additional 6.6 million pallets are therefore estimated to require repair at grocery manufacturers, based on the 18.28 percent repair frequency. Thus, the total quantity of pallets requiring repair in 1986 over the entire grocery distribution system is estimated at 17.2 million pallets.

The volume of wood raw material consumed as a result of grocery pallet repair also varies depending on the severity of damage sustained 165

by the pallet. Based on information provided by industry sources, an

upper limit can be placed on the volume of wood raw material used in the

repair of a grocery pallet. As noted earlier, any pallet that requires

more than the repair of four deck boards and one full—length stringer is

discarded. The volume of wood contained in the four deck boards and one

full-length stringer is about one-third of the total board foot volume

in a typical grocery pallet, which is considered to have three stringers

and 12 deck boards. Therefore, the upper limit on the volume of wood

raw material consumed in the production of pallet parts, ie., stringers and deck boards used in the repair of a grocery pallet, is 6.5 board foot per pallet, or one-third of the 19.5 board foot estimated for production of parts used in construction of a new grocery pallet. The total volume of wood raw material required for repair of grocery pallets thus has an upper limit of nearly 112 million board foot of additional wood raw material required in 1986. In combining the wood raw material volume estimated in 1986 for new grocery pallet production with the estimate of wood raw material for pallet repair, the total volume has the potential to be 21 percent of total hardwood consumption for all types of pallets in 1986, or over 950 million board foot of hardwood raw material.

The lower limit on the volume of wood raw material used in the repair of grocery pallets considers repair of one deck board, comprising approximately 1.5 board foot of wood raw material, as the minimum amount of repair incurred. This lower limit equals about 26 million board foot 166

of wood raw material for repair of grocery pallets in 1986. The average

volume of wood raw material consumed in the production of pallet parts

for repair of grocery pallets must fall between 6.5 and 1.5 board foot

of wood raw material per pallet, but a more precise estimate of this

volume is not possible given the available information. For this

reason, in distributing the volume of wood raw material required for

repair of pallets on a regional basis, I present the volumes required as

upper and lower estimates (Table 11).

( ( 167

Table 11. Estimated Wood Raw Material Consumption in

Grocery Pallet Repair, 1986.

UPPER LOWER

REGION LIMIT LIMIT

(MMBF) (MMBF)

New England 5.57 1.29

M16-Ac1am;1c 19 . 73 ls . 58 E South East 22.57 5.2N

East Central lü.57 3.38

West Central 18.ü7 N.29

South West 11.ü7 2.66

Pacific 19.62 N.56

TOTAL 112.00 26.00 168

INDUSTRY TRENDS IN WOOD USE

The grocery distribution industry has shown consistent growth, exceeding 5 percent annually, in the volume of products moved over the past 5 years (Progressive Grocers Information Sales 1987). This growth has been projected to continue over the next decade (Food Marketing

Institute 1988). The impact of continued growth in the grocery distribution industry on the wood resource must be considered in relation to several trends in product movement and pallet use observed at distribution centers or identified by distribution center managers.

These trends include the continued use of pallets in grocery distribution, continued preference for hardwood species in pallet construction, greater standardization in pallet size and construction characteristics, and potentially greater numbers of captive pallet operations with fewer exchanges of pallets between grocery manufacturers and distribution centers.

Continued Pallet Use

The demand for reusable pallets by the grocery and related products industries is expected to continue to be an important part in the overall demand for new pallets. None of the distribution center managers contacted in this study expect to replace the use of pallets in grocery distribution in the foreseeable future with alternative handling systems. The benefits derived from palletized handling of grocery and related products have been enumerated earlier. Managers expect these 169

benefits to continue through future use of pallets in handling grocery

products. As the volume of grocery products increases each year,

continued use of pallets in handling grocery products will increase the

demand for grocery pallets.

Continued Preference for Hardwood Species

Even though grocery pallets produced in 1986 only constituted

about 11 percent of the total number of new pallets produced, the volume

of wood raw material consumed in production of these pallets exceeded 18

percent of the total volume of hardwood raw material attributed to

pallet production. Based on information provided by distribution center

managers, the preference for hardwood species as the source of raw

material in construction of grocery pallets is consistent across all

regions. This preference appears to be based on the distribution center

managers' perceptions of a superiority of hardwood pallets over softwood pallets with regard to pallet durability, load—carrying capacity, and

overall length of expected service. Additionally, grocery pallet users show a reluctance in specifying higher cost, alternate materials for construction of pallets, that is other than hardwood lumber, as long as

they can continue to purchase low-cost hardwood lumber pallets in sufficient quantity to satisfy their needs. Thus, the cost of pallets made from hardwood raw material also plays a role in the continued preference for hardwood species in grocery pallet construction. 170

The effect on the hardwood resource in any particular region

resulting from changes in demand for reusable pallets by the grocery and

related products industries could be considerably different than the

effect resulting from changes in overall pallet demand. Although

pallets are produced in nearly every state, pallet manufacturers are

concentrated in the East Central and South East regions. And, the

pallet manufacturers in the East Central and South East regions are the

largest users of hardwood materials. As noted earlier, pallet

production tends to be located in areas where food processing or

distribution facilities are located. In Table 10, the breakdown of

regional demand for wood raw material in grocery pallet construction

shows that 72 percent of the wood raw material consumed in grocery

pallet production is used by producers in the combined Central and

Eastern regions. Because these regions also contain the bulk of the

nation's hardwood resources, the impact of increased growth in grocery

pallet use on this resource is likely to be less critical than it would

be in the Pacific or South West regions where hardwood resources are

less plentiful.

Increasing Standardization of Pallets

The trend toward even greater standardization in pallet size and construction characteristics than already exists in the industry is documented in recently released guidelines by the Joint Grocery Industry

Committee (Food Marketing Institute 1988). These guidelines are aimed 171

at reducing grocery product losses from shipping damage which are

estimated to be $2.5 billion annually. One of the industry

recommendations identified in these guidelines is the universal use of

standard ü8xü0 pallets. This recommendation is based on industry

estimates that 2 out of every 3 pallets currently in use do not conform

to industry guidelines for the N8xN0 pallet and that 20 percent of all

product damage can be attributed to sub-standard or damaged pallets.

The Grocery Pallet Council, formed in 1972 to manage pallet pool

operations in the food industry, developed a program for certifying that

pallets in the pool met specified construction characteristics (Wallin

198ü). This program was discontinued in 197U, but the GMA (Grocery

Manufacturers of America) specifications for pallet construction which

came out of this program have remained as unofficial guidelines for the

construction of grocery pallets. The new guidelines by the Joint

Grocery Industry Committee appear to be a step forward in the direction

of a new industry standard. The problem in implementing any industry

standard, as always, is the cost of a pallet which meets the standard as

opposed to the cost of a pallet which almost meets the standard.

Any grocery manufacturer or distribution center that purchases

standard pallets and exchanges them on a regular basis with other grocery pallet users must constantly be on guard to ensure that pallets conforming to the same standard are exchanged. If a sub-standard pallet can be purchased for a few cents less than the cost of a standard pallet and subsequently is exchanged for a standard pallet, the purchaser of 172

the sub-standard pallet has obtained a better quality pallet at a lower

cost. Over the course of a year, the money saved through purchasing

sub-standard pallets can amount to thousands of dollars for a typical

distribution center. However, these savings are illusions when the

sub-standard pallets are a direct contributor to grocery product damage

and loss.

Efforts by industry groups to promote standardization must

consider pallet construction characteristics in relation to overall

cost. Historically, pallet construction characteristics have changed

only in response to changes in the pallet users detailed specifications

for their pallets (Wallin 1977). However, the Joint Grocery Industry

Committee appears to be promoting more awareness of design procedures

which incorporate strength properties of wood and properties of

fasteners (Food Marketing Institute 1988). Thus, more efficient pallets

with satisfactory performance characteristics can be produced at a cost

that will reduce the incentive to purchase a pallet which does not

conform to the standard. This means that long-term growth of the

grocery pallet industry will not necessarily result in a proportional

increase in wood raw material use. The new design procedures and better

fastening devices may result in even less wood raw material used per pallet, which would result in a smaller proportional increase in wood

raw material use than would be expected on the basis of increased growth

in the volume of grocery products moved. 173

Captiva Pallet Operations

Several distribution center managers contend that a major trend in

future grocery pallet use is the possibility of more captive pallet

operations. In these operations, the user maintains control of the

pallet at all times and does not exchange the pallets with other users.

This trend has the potential to reduce the rate of growth in the volume

of wood raw material used in grocery pallet construction for two

reasons: increased longevity of pallet life resulting from closer

control of pallets in the system will reduce the frequency of new pallet

purchase; and, reduced occurrence of pallets with sub-standard

construction specifications will also reduce the frequency of pallet

damage requiring repair. The trend will not eliminate the use of

grocery pallets in grocery distribution. Thus, increases in the volume

of grocery products moved and stored will still result in increases, at

a reduced rate, in the volume of wood raw material used in grocery

pallet construction.

Because pallet users have more control over the captive pallets in

the system, they can expect that replacement or repair of sub-standard pallets will be reduced. One example of a captive pallet operation in a

distribution center warehouse illustrates this point dramatically: a

frozen food warehouse operated by Giant Food, Inc., in Jessup, Md., uses captive pallets in a completely automated, high-rise storage system.

The distribution center purchased 6,300 pallets for use in the system when it was constructed in 1981. Virtually all of the original 6,300 17'4

pallets remain in the system today with little visible sign of wear on

the pallets.

While the above example is exceptional in the longevity achieved

in individual pallet use, the opportunity of extending pallet life and

reducing the cost of repair or replacing pallets exists in other areas

of grocery distribution as well. The interest in captive pallet

operations is also linked with distribution center managers' observation of a trend over the past two years increasing the use of slip sheets in movement of products from manufacturers to distribution centers. At present, the industry average for shipping on slip sheets is less than

30 percent of total grocery products moved from manufacturers to distribution centers. The rate of increase in slip sheet use could not be identified from the response of distribution center managers; but, their responses clearly indicated that a continuation of the trend, if it actually existed, would permit closer control of pallets moving through the distribution center. While the distribution center would continue to handle the product on pallets from the receiving dock through final delivery to retail stores, the opportunity to eliminate the direct exchange of pallets with grocery manufacturers provides an opening for more captive pallet operations between distribution centers and retail stores. Again, only that portion of grocery pallet production currently going to grocery manufacturers would have an effect on the demand for wood raw material in grocery pallet production. CHAPTER VIII

SUMMARY AND CONCLUSIONS

The purpose of this chapter is to summarize the findings of

previous chapters and to discuss the implications of these findings.

Future research needs in the area of pallet use by the grocery and

related products industry are discussed in the concluding section.

SUMMARY

Since 1960, pallet production has quadrupled, increasing the

pallet industry's use of hardwood lumber from lü percent to almost 50

percent of total hardwood lumber production. Part of this growth can be

attributed to the grocery and related products industry, which should

continue as a potential major growth area for pallet usage over the next

decade. Reported growth in retail sales dollar volume of grocery and

related products, measured in real dollars, has averaged about 2.5

percent fro~ 1980 to 1986. Projected growth in manufacturer's production in the grocery products industry, measured in real dollar volume of products shipped, is expected to average 2.8 percent per year between 1986 and 1995.

If present pallet use and construction methods are continued, the resulting demand for pallets by the grocery and related products industry could have an uneven effect on the availability and price of the hardwood resource in the various regions of the country. Some

175 176

regions of the country have large quantities of currently underutilized » hardwood resources. These resources are available to meet the needs

ofincreasing demand for pallet raw material. Because raw material costs

are a substantial portion of the productlon cost of pallets, the regions

which can provide the lower cost raw material have a decided advantage

in future pallet production.

Information is needed for pallet producers to understand the

long—term potential and long—term trends in the grocery pallet market.

Also, in order to assess the regional impact that grocery pallet

production will have on future hardwood resources, better information is

needed on the current use of reusable pallets by the grocery and related

products industry. The pallets used in the distribution of grocery and

related products are believed to be unique in two ways: they are almost

exclusively made up of N8xüO size pallets, and are used consistently

within the industry to move and store products in preference to other

handling devices.

The general objective of this study is to provide information that

can be used to understand the long-term potential and long—term trends

in the grocery pallet ~arket as they relate to future regional timber

demands by the pallet industry. The specific objectives were:

(A) Provide information on current use of grocery pallets in the

grocery distribution industry through the identification and

quantification of grocery pallet use within the retailing and

wholesaling sectors of the grocery and related products industry. 177

(B) Provide a theoretical framework for future analysis of the

regional demand for grocery pallets resulting from the use of grocery

pallets in satisfying grocery distribution between regions and between

market areas within the same region, and determine the relationship

between grocery pallet use and regional grocery pallet demand under

specific food flow and pallet durability assumptions.

(C) Provide information on the demand for regional timber

resources resulting from grocery pallet production within specified

regions.

Objective A was fulfilled by the specification and documentation

of market models for the grocery and related products market and the

grocery pallet market. These market models consisted of four functional

relationships representing aggregate demand and supply in the grocery

and related products market, and an additional four functional

relationships representing the aggregate demand and supply in the

grocery pallet market. In the grocery pallet market model, the demand

for new grocery pallets was expressed as an 'excess demand' where the quantity of new grocery pallets demanded equaled the difference between

the total quantity of pallets used in the movement or storage of grocery and related products in a given time period and the available inventory of grocery pallets in that time period. The available inventory of grocery pallets in the grocery distribution system was expressed as a function of the dollar volume of retail sales, based on an application of a stock adjustment model for a durable input. 178

Objective B was fulfilled by the specification of an equation

representing the quantity of new grocery pallets consumed in the grocery

distribution system in terms of the inventory quantity of grocery

pallets in the system. The inventory quantity of pallets in the system

was expressed in terms of the dollar volume of retail sales reported in

each region. Regional pallet consumption estimates were made subject to

the qualification that the regional allocation of pallet consumption by

grocery manufacturers was not exact.

Based on the estimates of regional grocery pallet consumption, the

volume of wood raw material consumed in production of new grocery

pallets and repair of damaged grocery pallets was estimated for

specified regions. Because grocery pallet consumption was estimated

rather than grocery pallet demand, the price—quantity relationships that

could be used to measure the effect of changes in grocery pallet demand

on the timber resource were not determined. These price-quantity

relationships could be determined in future estimation of the demand and

supply model for grocery pallets, provided that additional grocery

pallet price and quantity data became available. National trends

affecting wood use in grocery distribution were considered. However, specific regional trends affecting wood raw material use were not identified.

In documenting the market models, the spatial aspects of the grocery distribution industry were discussed along with the perceived flow of pallets through the distribution system. Because of the 179

tendency for food production, distribution, and consumption to be

spatially separated activities in the grocery and related products

industry, each of these activities required movement of the grocery and

related products. This movement was shown to be satisfied, in large

part, by the use of pallets. The distinction made between pallet demand

and supply points and the movement of pallets between grocery demand and

supply points was a key element in the discussion of the spatial aspects

of grocery distribution. V

Pallet supply points were identified by the location of pallet

manufacturers that produced grocery pallets. Although 2,u70 pallet

manufacturers were reported in the 1985 survey, the number of pallet

manufacturers actually producing grocery pallets was not available but was considered to be less the the total number of pallet manufacturers

in the industry. The market for grocery pallets was considered to be highly competitive primarily because pallet manufacturers have considerable ease of entry and exit in the market for grocery pallets.

Pallet demand points were identified by the location of grocery manufacturers and distribution centers that use grocery pallets in the movement and storage of grocery and related products. The #09 distribution centers identified in the study were all considered to be pallet demand points, in that they all used pallets at least for the storage and movement of grocery products within the distribution center warehouse. In addition, the location of these distribution centers was specified by region and by market area within a region. On the other 180 hand, the 22,130 grocery manufacturers were identified only as potential pallet demand points. The percent of grocery manufacturers actually using pallets was an unknown in this study. Also, the location of grocery manufacturers using grocery pallets was not identified as specifically as the location of the distribution centers. This factor created a major obstacle in subsequent identification of regional pallet demand. .

The flow or movement of grocery pallets through the distribution system was shown to revolve around and through individual grocery distribution centers. However, this movement of pallets depended on the demand for grocery products at retail stores and at distribution centers as well as on the supply of grocery products from grocery manufacturers to distribution centers and from distribution centers to retail stores.

The movement of grocery and related products from grocery supply points to grocery demand points generated the demand for grocery pallets at each point of shipment origin. Had this demand for grocery pallets been satisfied by supply of grocery pallets from pallet manufacturers each time the demand occurred, the market models would have been simplified.

However, grocery pallets were shown to be a durable input to the production and distribution of grocery products. That is, an inventory quantity of pallets was shown to exist in the grocery distribution system. While an aggregate level of demand for grocery pallets resulted from the movement and storage of grocery products, the demand for new grocery pallets supplied by pallet manufacturers was shown to be a 181

function of the quantity of pallets available in the system and the

overall level of aggregate pallet demand. The circular nature of pallet

flows in the system was indicated simply because grocery pallets were

recycled through the system as long as they remained in serviceable

condition.

Pallet repair operations were included in the description of

pallet flows in grocery distribution. However, pallet repair, whether

handled in-house or contracted out, was considered as a maintenance

operation and as such, entered into grocery manufacturers' and

distribution centers' production functions as a fixed cost.

Neither the quantity of new grocery pallets nor the total

inventory quantity of pallets in the grocery distribution system could

be identified from published data. Therefore, estimation of the

quantity of new pallets and the inventory quantity of pallets in the

system was made based on assumptions about the relation of retail sales

to quantity of pallets in inventory. The accuracy of these assumptions was dependent on the accuracy of the responses received in interviews with distribution center managers. In a number of these interviews, the

responses were considered to be ”best guess" estimates of existing

industry conditions on the part of distribution center managers. For

this reason, considerable variation was expected between the estimated quantity and actual quantity of pallets, either in inventory or as new pallets entering the system. The only reference points for validating the estimate of M3 million new grocery pallets purchased in 1986 were 182

three surveys conducted in 1979, 1982, and 198#. The estimate of new

grocery pallets purchased in 1986 did fall within the range expected

fro~ the information provided in the three surveys.

The 1986 estimate of new grocery pallets purchased nationally was

broken down into regional estimates of new pallet consumption. In

making these regional estimates, it was noted that only 62 percent of

the new pallet consumption could be expected to occur at distribution

centers while the remaining 38 percent was more likely to be attributed

to grocery and related products manufacturers. Since the grocery

manufacturing facilities could not be shown to be located in the same

regions as the distribution centers they served, the regional estimates

of new pallet consumption were considered primarily to reflect the

consumption or purchase of new pallets by distribution centers.

The volume of wood raw material used in 1986 for production of

grocery pallets was estimated to exceed 838 million board foot of hardwood raw material, or potentially over 18 percent of the #.5 billion board foot of hardwood raw material consumed in the production of all

types of pallets. Regional allocation of this volume of hardwood raw material was also considered to reflect primarily the consumption of hardwood raw material for production of pallets by distribution centers.

The volume of wood raw material used in the repair of grocery pallets in 1986 was expressed in terms of an upper and lower limit to l the volume of wood raw material used in repair operations. At least 26 million board foot and no more than 112 million board foot of wood raw 183

material was considered to be required for the repair of grocery pallets

in 1986. If the volume of hardwood raw material actually used in repair

of grocery pallets approached the upper limit of 112 board foot, the

estimated total volume of hardwood raw material consumed in the

production of pallet parts in 1986, estimated to exceed 950 million board foot, was shown to potentially exceed 20 percent of the total hardwood consumption for all types of pallets.

General industry trends in wood use were considered to include continued use of hardwood pallets in grocery distribution, marked preference for hardwood species over softwood species in pallet construction, more emphasis on standardization in terms of pallet quality and physical characteristics, and a potential for more captive pallet operations in the future.

CONCLUSIONS

The general objective of this study was to provide information about the economic structure of the grocery distribution industry. In accomplishing this objective, I have provided a framework, based on economic theory, that can be built on as additional data become available for empirical estimation of the functional relationships comprising the market models. Potential or future changes in the grocery pallet industry cannot be estimated empirically given the present availability of economic data. In order to estimate the econometric models specified in Chapter 5, grocery pallet price and 18ll

quantity data must be collected either for specified regions or

individual market areas as outlined in Chapter 6.

A stock adjustment model was used to develop a functional

relationship between grocery pallet inventory levels and the volume of

retail sales. Application of the stock adjustment model in this study

of a demand for a durable input was not unique. Other demands for

durable inputs have been examined using this tool. However, in dealing

with the problem of pallets remaining in the system from one time period

to another, the stock adjustment model provided a logical and consistent

technique for bridging the gap between the quantity of pallets

considered in the grocery manufacturers and distribution centers

production functions and the demand function for new purchases of

pallets.

The consumption of grocery pallets by the grocery distribution

industry was shown to be an important part in the overall new pallet

production even though the pallet used, N8xüO inches, only constituted

about 11 percent of the total number of new pallets produced in 1986. A

disproportional effect on the hardwood resource was shown for grocery

pallets in relation to the quantity of grocery pallets consumed. With

Q over 20 percent of the total hardwood consumption for all types of

pallets being attributed to the manufacture of grocery pallet parts, the

effect on the hardwood resource in any particular region resulting from

changes in the grocery pallet consumption can potentially be greater

than the effect resulting from changes in consumption of other pallet 185

types, particularly those using softwood raw material in pallet

production. However, until more precise estimates of the regional

production levels for pallets of all types become available, such a

comparision cannot be quantified.

Grocery pallet production was shown to be located in areas where

food manufacturing and distribution facilities are located. Because 72 percent of the wood raw material required by the grocery pallet industry

in 1986 was esti~ated to be consumed in the Central and Eastern regions, the regional estimates of hardwood raw material consumed in grocery pallet construction appear to reflect reported actual consumption.

Census data for 1982 indicated that the pallet industries in the Central and Eastern regions were in fact the largest users of hardwood materials.

Grocery pallet users will be reluctant to specify higher cost, alternate materials for construction of grocery pallets as long as they can continue to purchase low-cost pallets made from hardwood lumber in sufficient quantity to satisfy their needs. Continued use of hardwood pallets can be expected in grocery distribution for some time.

However, long-term growth in the demand for grocery pallets will not necessarily result in a proportional increase in wood raw material use. As pallet manufacturers begin to make more use of design procedures incorporating strength properties of wood and properties of fasteners, more efficient pallets with satisfactory performance characteristics will be produced. These new design procedures and 186

better fastening devices will in all probability result in even less

wood raw material used per pallet. Thus, increases in the volume of

wood raw material used in grocery pallet production will still depend on

continued increases in the volume of grocery and related products moved

and stored on pallets, but the rate of increase for wood raw material

consumption may be less than the rate of increase in the volume of

products ~oved.

The recent renewal of industry interest in increased

standardization should result in a better quality pallet in use in the

future. As the quality of pallets improves, increased service life and

reduced rates of repair should be expected. Both of these factors would

have the effect of further reducing the rate of growth in the demand on

the timber resource by grocery pallets in the future.

FUTURE RESEARCH

Limited data availability for estimation of the models presented

suggests that one area for future research includes the development of

additional sources of data relating to regional production of pallets,

specified in terms of price, quantity, and dimensions as well as

expected end use. Other data items needed to estimate the stock

adjustment model include dollar volume of grocery and related products

retail sales either for specific regions or for individual market areas over an extended time period rather than just one year as presented in this study. Institutions such as Progressive Grocers Information Sales, 187

Inc., or the Food Marketing Institute are the most likely candidates for

future collection of data regarding dollar volume of grocery and related products retail sales. The collection of the above data could be a part of a broader data collection effort on the part of a continuing U.S.

Forest Service research effort aimed at identifying both the present and future regional use of the hardwood resource.

Having acquired the necessary price and quantity information relating to grocery pallets, the second priority for any future research should be the actual estimation of the demand and supply models presented in this study. Until these or similar models are estimated,

I cannot quantify the effect of changes in grocery pallet demand on the timber resource in the different regions of the country. Specific benefits of having these data and being able to estimate the conceptual models formulated earlier are dependent on how critical the information provided by estimation of the models is to development of long-term forest policy and to understanding the long-term potential in the grocery pallet segment of the overall pallet industry. The impact of such information remains to be determined ln future research.

Measuring the effect of changes in grocery pallet demand on the timber resource was an objective in this study that could not be reached, given the absence of estimated grocery pallet demand and supply relationships. Therefore, the next priority for any future research after the esti~ation of grocery pallet demand and supply relationships should be the application of these relationships in a more detailed 188

analysis of the wood raw material demand derived from the production of

grocery pallets.

Another opportunity for future research lies in developing more

complete information on the extent and location of grocery pallet use in

the manufacturing segment of the grocery and related products industry.

Although this was considered to be outside the scope of the present

study, on further reflection it would have made a considerable

difference in the accuracy of the regional estimates of grocery pallet

consumption.

Finally, the importance of the relationship between price of new

pallets and the price of repaired pallets has been discussed in

theoretical terms in this study. However, a more exact estimation of

the effect of changes in the price of new grocery pallets on the

quantity of grocery pallets repaired remains to be determined by future

research.

I will conclude the discussion of future research by reiterating

the statement that the impact of continuing research in this area on long-term forest policy and better understanding of the long-term potential in the grocery pallet market cannot be known until more information is made available. For this reason, I believe that additional research in this area is required until such time when a definite statement regarding the impact of developing that information can be made. LITERATURE REFERENCES

Adams, D. M. 1974. Forest Products Prices and National Forest Timber Supply in the Douglas-fir Region. Forest Science. 20(3): 243-259.

Adams, D. M.; Haynes, R. W. 1980. The 1980 Softwood Timber Assessment Market Model: Structure, Projections, and Policy Simulations. Forest Science Monograph No. 22. Forest Science. 64 p.

Adams, F. G.; Blackwell, J. 1973. An Econometric Model of the United States Forest Products Industry. Forest Science. 19(2): 82-96.

Anderson, R.B. 1986. Future Availability of Pallet Raw Material in the South. Pallet Enterprise. 6(6): 44-46.

Anderson, R.B. 1987. Future Availability of Pallet Raw Material in the North. Pallet Enterprise. 7(3): 31-34.

Arzac, E.R.;Wilkinson, M. 1979. A Quarterly Econometric Model of United·States Livestock and Feed Grain Markets and Some of Its Policy Implications. Am. J. Ag. Econ. 61(2): 297-308.

Beilock, R,;Casavant, K. 1984. Perishables - The New Intermodal Battleground. Am. J. Ag. Econ. 66(5): 651-656.

Beilock, R.;Freeman. J. 1984. Florida Motor Carrier Deregulationz Perspectives of Urban and Rural Shipper/Receivers. Am. J. Ag. Econ. 66(1): 91-98.

Berkovec,J. 1985. New Car Sales and Used Car Stocks: A Model of the Automobile Market. RAND Journal of Economics. 16(2): 195-214.

Bigman, David;Reutlinger, Shlomo. 1979. Food Price and Supply Stabilization: National Buffer Stocks and Trade Policies. Am. J. Ag. Econ. 61(4): 657-667.

Blackie, Malcolm J.;Dent, J.B. 1976. Analyzing Hog Production Strategies with a Simulation Model. Am. J. Ag. Econ. 58(1): 39-46.

Bobst, B. W.;Waananen, M. V. 1968. Cost and Price Effects of Concentration Restrictions in the Plant Locatian Problem. Am. J. Ag. Econ. 50(3): 676-686.

189 190

Brandow, G. E. 1969. Market Power and Its Sources in the Food Industry. Am. J. Ag. Econ. 51(1): 1-12.

Buongiorno, Joseph;Gilless, J. K. 1981. Concepts Used in a Regionalized Model of Pulp and Paper Production and Trade. Forest Sector Models: Proceedings of the First North American Conference: International Institute for Applied Systems Analysis: Williamsburg, Va. 57-70.

Buongiorno, Josephgüilless, J. K. 1983. Modeling the North American Paper Industry, with an extension to International Trade. Proceedings of the Second Annual Meeting: International Institute for Applied Systems Analysis: Washington, D.C. 117-133.

Buongiorno, Joseph;Gilless, J. K. 1983. A Model of International Trade of Forest Products (GTM-1). WP-83-63, International Institute for Applied Systems Analysis: Laxenburg, Austria. 2üp.

Byrkett, D.L.:Miller, R.A.;Taiganides, E.P. 1976. Modeling the Optimal Location of the Cattle Feeding Industry. Am. J. Ag. Econ. 58(2): 236-2üN.

Chow, G. C. 1957. Demand For Automobiles in the United States. North-Holland Publishing Company, Amsterdam. 110 p.

Chow, G. C. 1960. Statistical Demand Functions for Automobiles and Their Use For Forecasting. In: The Demand For Durable Goods. ed. Arnold C. Harberger. Univ. of Chicago Press: Chicago, Ill. 1N9-178.

Conner, J.;et al. 1985. Economic Forces Shaping the Food-processing Industry. Am. J. Ag. Econ. 67(5): 1136-11ü2.

Duloy, John H.:Norton, Roger D. 1975. Prices and Incomes in Linear Programming Models. Am. J. Ag. Econ. 57(N): 591-600.

Duncan, R. S. 1980. The Effect of Gasoline Prices on Automobile Sales. Am. Econ. 1ü(1): 62-66.

Edwards, R.;Parikh, A. 1976. A Stochastic Policy Simulation of the World Coffee Economy. Am. J. Ag. Econ. 58(2): 152-160.

Eichler, John R. 1976. Wood pallet manufacturing practices. Eichler Associates: Cape Coral, Fla. 277 p.

Emanuel, D. M. 1985. Wooden Pallet Manufacturers. Unpublished Report. U. S. Forest Service, Princeton, W. Va. 115 p. 191

Fedeler, J. A.:Heady, E. 0. 1976. Grain Marketing and Transportation Interdependencies: A National Model. Am. J. Ag. Econ. 58(2): 22U-235.

Food Marketing Institute. 1988. Voluntary Industry Guidelines for Dry Grocery Shipping Containers. Food Marketing Institute: Washington, D.C. 16p.

Fromm, Gary. 1973. Implications to and from Economic Theory in Models of Complex Systems. Am. J. Ag. Econ. 55(2): 259-271.

Fuchs,H.W.;Farrish,R.0.P.;Boha1l,R.W. 197ü. A Model of the U.S. Apple Industry: A Quadratic Interregional Intertemporal Activity Analysis Formulation. Am. J. Ag. Econ. 56(N): 739-750.

Furtan, W.H.:Nagy, J.G.;Storey, G.G. 1979. The Impact on the Canadian Rapeseed Industry from Changes in Transport and Tariff Rates. Am. J. Ag. Econ. 61(2): 238-2ß8.

Gemmill, Gordon. 1977. An Equilibrium Analysis of U. S. Sugar Policy. Am. J. Ag. Econ. 59(N): 609-618.

Goehring, C. B.: Wallin, W. B. 1980. A survey of loads, loading conditions, and handling environment conditions for wooden pallets. Unpubl. Report. U. S. Forest Service, Northeastern Forest Experiment Station: Princeton, W. Va. 13 p.

Gordon, R. J. 1978. Macroeconomics. Little, Brown and Company: Boston, Mass. 585 p.

Griliches, Z. 1960. The Demand For A Durable Input: Farm Tractors in the United States, 1921-1957. In: The Demand For Durable Goods, ed. Arnold C. Harberger. Univ. of Chicago Press: Chicago, Ill. 181-207.

Handy, C. R.;Padberg, D. I. 1971. A Model of Competitive Behavior in Food Industries. Am. J. Ag. Econ. 53(Z): 182-190.

Hassan, A. A.; Wisdom, H. W. 1982. International Trade Models for Selected Paper and Paperboard Products. Department of Forestry, Virginia Polytechnic and State University: Blacksburg, Va.

Hazell, P.B.R. 1979. Endogenous Input Prices in Linear Programming Models. Am. J. Ag. Econ. 61(3): H76-#81.

Heady, Earl 0.;Hall, Harry H. 1968. Linear and Nonlinear Spatial Models in Agricultural Competition, Land Use, and Production Potential. Am. J. Ag. Econ. 50(5): 1539—15Ü8. 192

Heien, D. M. 1980. Markup Pricing in a Dynamic Model of the Food Industry. Am. J. Ag. Econ. 62(1): 10-18.

Houthakker, H. S.; Taylor, L. D. 1966. Consumer Demand in the United States, 1929-1970. Harvard University Press: Cambridge, Mass. 214 p.

Houthakker, H. S.; Taylor, L. D. 1970. Consumer Demand in the United States: Analyses and Projections. Harvard University Press: Cambridge, Mass. 321 p.

Huang, D. S. 1964. Discrete Stock Adjustment: The Case of Demand for Automobiles. International Economic Review. 5(1): 46-62.

Johnson, P.R.;Grennes, T.;Thursby, M. 1979. Trade Models with Differentiated Products. Am. J. Ag. Econ. 61(1): 120-127.

Kaylin, S. 0. 1968. Understanding Today's Food Warehouse. Chain Store Publishing Corp.: New York, New York. 226p.

Kilmer, R.L.;Spreen, T.;Ti11ey,D.S. 1983. A Dynamic Plant Location Model: The East Florida Fresh Citrus Packing Industry. Am. J. Ag. Econ. 65(4): 730-737.

King, Richard A.; Ho, Foo-Shiung. 1972. Reactive Programming: A Market Simulating Spatial Equilibrium Algorithm. Economics Research Report No. 21. Department of Economics, North Carolina State University: Raleigh, North Carolina. 48 p.

Kloth, Donald W.;Blakely, Leo V. 1971. Optimum Dairy Plant Location with Economies of Size and Market-Share Restrictions. Am. J. Ag. Econ. 53(3): 461-466.

Kmenta, J. 1971. Elements of Econometrics. MacMillan Publishing Company: New York, New York. 655 p.

Kottke, Marvin. 1970. Spatial, Temporal and Product-Use Allocation of Milk in an Imperfectly Competitive Dairy Industry. Am. J. Ag. Econ. 52(1): 33-40.

Labys, W. C.; Pollak, P. K. 1984. Commodity Models for Forecasting and Policy Analysis. Croom Helm Commodity Series, Croom Helm Ltd.: London, England. 48-102.

Ladd, George W.;Lifferth, Dennis R. 1975. An Analysis of Alternative Grain Distribution Systems. Am. J. Ag. Econ. 57(3): 420-430. 193

Lamm, R. McFall, Jr. 1976. Market Performance in the U.S. Vegetable Oil Industry: An Application of Simulation for Market Analysis. Am. J. Ag. Econ. 58(#): 735-739.

Lamm, R. McFall, Jr.;Westcott, P.C. 1981. The Effects of Changing Input Costs on Food Prices. Am. J. Ag. Econ. 63(2): 187-196.

Lancaster, K. J. 1966. A New Approach to Consumer Theory. Journal of Political Economy. 7#(#): 132-157._

Libbin, J. D.;Boehlje, M. D. 1977. Interregional Structure of the U. S. Coal Economy. Am. J. Ag. Econ. 59(3): #56-#66. H.1 Manning, G. 1975. The Canadian Softwood Lumber Industry: A Model. Canadian Journal of Forest Research. 5(3): $#5-$51.

Martin,Larry;Zwart,Anthony C. 1975. A Spatial and Temporal Model of the North American Pork Sector for the Evaluation of Policy Alternatives. Am. J. Ag. Econ. 57(1): 55-66.

McCurdy, D. R.; Ewers, J. T. 1985. The U. S. Pallet Industry. Pallet Enterprise. #(6): 8-12.

McCurdy, D. R.; Ewers, J. T. 1986. The Pallet Industry in the United States 1980 and 1985. Dept. of Forestry, Southern Illinois University: Carbondale, Ill. 13 p.

McCurdy, D. R.; Ewers, J. T.: Kung, F. H.; McKeever, D. B. 1988. A Study of Lumber Use in Pallets Manufactured in the United States: 1982 and 1985. Forest Products Journal. $8(2): 11-15.

McKeever, D. B.; Dickerhoof, H. E. 1980. Lumber and panel products consumption for packaging and shipping in the United States - perspective for the 1980's. Resour. Bul. FPL-10. U. S. Forest Service, Forest Products Laboratory: Madison, Wis. 5 p.

McKeever, D. B.;McCurdy, D. R.;Kung,F. H.;Ewers, J. T. 1986. Wood Used in Pallets Manufactured in the United States, 1982. Resour. Bul. FPL-RB-17. U. S. Forest Service, Forest Products Laboratory: Madison, Wis. 1$p.

McKillop, W. L. M. 1967. Supply and Demand for Forest Products--An Econometric Study. Hilgardia. $8(1): 132 p.

Miklius, W.:Casavant, K.L.;Garrod, P.V. 1976. Estimation of Demand for Transportation of Agricultural Commodities. Am. J. Ag. Econ. 58(2): 217-223. . 19N

Monke, E.;Petzel, T. 198N. Market Integration: An Application to International Trade In Cotton. Am. J. Ag. Econ. 66(ü): M81-u87.

National Wooden Pallet and Container Association. 1980. Unpublished pallet industry survey data. National Wooden Pallet and Container Association: Washington, D.C.

National Wooden Pallet and Container Association. 1983. 1983 Production Report. National Wooden Pallet and Container Association: Washington, D.C. 1 p.

National Wooden Pallet and Container Association. 1985. Bulletin #85-26. National Wooden Pallet and Container Association: Washington, D.C. 1 p.

National Wooden Pallet and Container Association. 1986. 1986 Production Report (New Pallets). National Wooden Pallet and Container Association: Washington, D.C. 1 p.

Nichols, Jr., T. E. 1969. Transportation and Regional Development in Agriculture. Am. J. Ag. Econ. 51(5): 1ü55—1ü63.

Panshin, A. J.;Harrer, E. S.;Bethel, J. S.;and Baker, W. J. 1962. Forest Products Their Sources, Production, and Utilization. McGraw-Hill Book Company: New York, New York. 538p.

Parker, R.C.;Conner, J.M. 1979. Estimates of Consumer Loss Due to Monopoly in the U.S. Food-Manufacturing Industries. Am. J. Ag. Econ. 61(ü): 626-639.

Pearce, D. K.;Wisley, T. 0. 1983. Sales Expectations and Inventory Changes in Retail Trade. Journal of Economics and Business. 35(1): 109-121.

Progressive Grocers Information Sales. 1987. 1987 Marketing Guidebook. Progressive Grocers Information Sales: Stamford, Connecticut. 809p.

Ryland, G.J.;Guise, J.W.B. 1975. A Spatiotemporal Quality Competition Model of the Australian Sugarcane Processing Industry. Am. J. Ag. Eccm. 57(3): u31—u38.

Samuelson, P. A. 1952. Spatial Price Equilibrium and Linear Programming. American Economic Review. ü2(6): 283-303. 195

Sendak, P. E. 1971. Market structure, conduct, and performance in the hardwood pallet industry. PhD dissertation. University of Massachusetts: Amherst, Mass.

Stollsteimer, J. F. 1963. A Working Model for Plant Numbers and Locations. J. Farm Economics. 95(8): 631-695.

Strobel, J. J.; Wallin, W. B. 1969. The unit-load explosion in the food industry. Res. Pap. NE-121. U. S. Forest Service, Northeastern Forest Experiment Station: Broomall, Pa. 60p.

Tramel, Thomas E. 1965. Reactive Programming——An Algorithm for Solving Spatial Equilibrium Problems. Technical Publication No. 9. Mississippi Agricultural Experiment Station: State College, Mississippi. 52p.

U.S. Department of Agriculture, Forest Service. 1982. An analysis of the timber situation in the United States 1952-2030. FS-23. U.S. Department of Agriculture, Forest Service: Washington, D.C. 999p.

U.S. Department of Commerce, Bureau of the Census. 1976. 1972 Census of Transportation. Vol. 3. U.S. Government Printing Office: Washington, D.C. 259p.

U.S. Department of Commerce, Bureau of the Census. 1981. 1977 Census — of Transportation. Commodity Transportation Survey Summary. U.S. Government Printing Office: Washington, D.C. 920p.

U.S. Department of Commerce, Bureau of the Census. 1981. 1977 Census of Manufacturers. U.S. Government Printing Office: Washington, D.C.

U.S. Department of Commerce, Bureau of the Census. 1985. 1982 Census of Manufacturers. U.S. Government Printing Office: Washington, D.C. von Oppen, Matthias;Scott, John T. 1976. A Spatial Equilibrium Model for Plant Location and Interregional Trade. Am. J. Ag. Econ. 58(3)= 937-995 - Wallin, W. B. 1989. Another Look At a National Pallet Pool. Pallet Enterprise. 9(1): 29-25, 28-30.

Wallin, W. B. Characteristics of the U.S. pallet industry. 1977. Unpubl. Report. U.S. Forest Service, Northeastern Forest Experiment Station: Princeton, W. Va. 79p. 196

Wallin, W. B.; Luppold, W. G. 1983. Impact of hardwood timber supplies on the pallet industry. Unpubl. Report. U.S. Forest Service, Northeastern Forest Experiment Station: Princeton, W. Va. 12p.

Walsh, J. 1981. Industry Study #P69--Industrial Packaging. Predicasts, Inc.: Cleveland, Ohio. 106p.

Warrack, Allan A.;Fletcher, Lehman B. 1970. Plant-Location Model Suboptimization for Large Problems. Am. J. Ag. Econ. 52(ü): 587-590• APPENDIX A

The following are commonly accepted definitions of terms used in the

grocery industry:

SUPERMARKET: A complete, departmentalized grocery store with minimum

annual sales of $2,000,000.

CHAIN: A company which operates eleven or more stores in total. A

chain store unit ls a store operated by such a company.

INDEPENDENT: A firm which operates from one to ten stores.

UNAFFILIATED INDEPENDENT: Operator from one to ten stores having no

affiliations with any organization and buying entirely from wholesalers

or suppliers on an independent basis.

COOPERATIVE: Independent grocers who jointly own and operate their own wholesale organization.

VOLUNTARY: A group of independent grocery stores jointly sponsored by

an independent wholesaler.

CONVENIENCE STORE: A small store handling a limited variety of items in general use in most households and patronized for fil1—in home requirements.

197 APPENDIX B

DERIVATION OF THE INVENTORY STOCK ADJUSTMENT MODEL

Net change in quantity of pallets in the system, after adjusting

for the effect of pallets that have been discarded, constitutes the

quantity of new pallets demanded in the grocery distribution system in

a given time period. This quantity, noted earlier in Chapter 5 as

QNP't, is expressed as a function of changes in the dollar volume of

retail sales for grocery and related products:

(1) ONPJ: = T { Xt · (1 · w*)Xt_1 } or, in terms of the quantity of pallets in the system: ‘ ‘ (2) °t (1 ****%-1 By rearranging the terms in this equation, we can express the

inventory quantity of pallets in the system in the current time period,

Qt, in terms of the demand for new pallets, QNP, and some portion of the inventory quantity of pallets in the preceeding time period, or ‘ ‘ **)%-1 °t °m> * (1 This relationship is assumed to hold for preceeding time periods so that we can express the quantity of pallets in the system in the previous time period, Qt_1, in terms of the demand for new pallets in that time period, and some portion of the quantity of QNP,t_1, pallets in the system in the preceeding time period, or ’ ’ "’*)°:-2 Qt—1 QNP,t—1 (1 ' Substituting this expression for in the above equation Qt Qt_1 for yields:

198 199

t—nth IExpandingthis expression to the time period yields:

·I· *2

·I> +(1-w)°Qt_n _ » e Since w is expressed as a positive percentage, the value of (1-w ) must be less than one. Thus, at the limit, in expansion of the above equation to include successively earlier levels of demand, the value of •

the last term, (1 - w )nQt_n, reduces to zero. This occurs because as the value of n approaches infinity, the fraction, (1 • - nth w ), raised to the power approaches zero. Thus, the inventory quantity of pallets in the system can be expressed as the sum of the demand for new pallets in the current time period and a decreasing portion of the demand for new pallets in all previous time periods. APPENDIX C

TYPICAL GROCERY PALLET SPECIFICATIONS

I. Size and Type of Pallet

Pallet shall be N8" x NO"; flush, non-reversible, four—way

modified.

II. Type and Quality of Lumber

Lumber shall be sound, square edge, free of mold, decay, and

noxious odors. The following hardwood species may be used for

stringers: Beech, Birch, Eucalyptus, Hackberry, Hard Maple, Hickory,

Oak (except Swamp Oak), Pecan, Rock Elm, White Ash. In addition to the

above list, the following hardwood species shall be used for

deckboards: Ash, Butternut, Magnolia, Red Alder, Soft Elm, Soft Maple,

Sweet Gum, Sycamore, Tupelo, Walnut, Yellow Poplar.

Any degree of seasoning is acceptable. ’ The diameter of sound knots shall be no greater than one-third

the width of the piece in which they occur; there shall be no more than

two such maximum diameter knots in any one piece. Loose or hollow knots shall not exceed one-half the diameter of sound knots.

No knots over 1/2" shall be allowed in the stringer immediately over the notched areas.

Length of crack or grain separation must be no longer than two-thirds the width of the piece in end deckboards and no longer than

200 201

twice the width of the piece in stringers and inside boards. Splits

running through full thickness of the piece (not to be confused with

nail splits) are permitted in any number, but when appearing in

endboards must be straddled by nails. Shakes are permitted in any

piece if contained by nailing.

Wane within limits is permitted on any piece provided it is not

on exposed edge of end deckboards. Wane may appear on surface or edge of other pieces but in no cases are nails to be driven into or through wane. No more than one-half of the pieces in an individual pallet may

contain wane.

No individual piece on any one pallet shall have deviation due to warp which is greater than the following percent of its measured — — dimension: bow 2%; crook - 2%; cup 3%.

Pin-worm and grub worm holes in pallet parts are permissible defects, providing that they do not affect the structural strength of the pallet. Infestation of wood-destroying insects is not permitted in pallet parts.

No combination of defects which will materially weaken any piece or pallet shall be allowed.

III. Lumber Preparation

Stringers and deckboards shall be smooth sawn or surfaced to square edge, uniform dimensions. 202

No bearded ends are permitted; all ends and edges shall be clean

and square.

Preservatives not permitted.

IV. Type and Quality of Fastener

Hardened steel screw nail (or tempered) , 2-1/ü" x 0.110", to

bend not more than 28 degrees on the MIBANT Nail Tester or equal. Wire

diameter 0.110”, threaded 0.D. body diameter 0.138", flat head nail 9/32”, diameter or snag free head average diameter 21/6ü". Point shall

be diamond (not longer than 5/32") or chisel provided that the width

does not exceed wire diameter. Helically threaded with four flutes;

helical angle of thread at the pitch diameter shall be 60 degrees plus

or minus 5 degrees with a plane perpendicular to the axis.

V. Dimensions, Spacing, and Arrangement

Deckboard dimensions shall be thickness: 13/16" to 15/16" range;

length: 39-7/8" to NO-1/8" range. 5-5/8” Extreme top end deckboards will be minimum width. All

intermediate top deckboards may be random widths, and will be minimum 3” 3-5/8”. Maximum spacing will be maintained. Deckboards must provide a minimum actual cumulative surface of 30".

Bottom end deckboards shall be 5-5/8" to 6" range in width. Two outside boards in the center cluster shall be minimum 3-5/8". The center board shall be minimum 5-5/8". Spacing of center cluster of 203

bottom deckboards shall not exceed 2-1/2" and the outer edges of the

cluster shall be flush with the inside of the notches. Bottom

deckboards shall not protrude over any notch opening; ends of all

deckboards shall be flush with outside of stringers. The center

cluster of deckboards shall provide a minimum actual cumulative surface

of 13".

Chamfer inner and outer edges of bottom deck edge boards and

wheel space edges of center cluster boards to within 1/2" of bottom

face. All chamfers to be cut at N5 degrees angle and to extend within

2" but not closer than l" of stringers at each end of chamfer.

Stringer dimensions shall be width: 1-3/N", plus 1/N", minus 0;

length: N8", plus or minus 1/8"; height: 3-3/N", plus 1/N", minus 0.

Notch openings shall be 9" wide, plus 1/N", minus 0. Notch

openings shall be no closer than 6" from the ends of the stringers,

plus 1/N", minus 0. Stringer notches shall have a depth of 1-1/2" and

shall have round corners with a radius of 3/N"; the top of the notch

shall be flat cut between the corner radii.

Center stringers shall be parallel to and equidistant between outside stringers.

VI. Assembly

Predrilling of deckboards shall be required when nails are hand driven. When pallets are assembled by nailing machines, predrilling of deckboards is not required. zou

The number of nails which shall be employed at all bearing points

for the various widths of deckboards is as follows: 3-5/8" to 5·1/2"

width 2 nails; 5-5/8" to 7-1/2" width - - 3 nails; 7-1/2" to 8" width - H nails.

Nails shall be staggered. Flat head nails shall be counter sunk

at least 1/16" deep. Oval concave "snag free" head nails shall be

driven flush with the deckboards.

VII. Workmanship

Protruding nail heads or points are not permitted. Bent over

nails must be driven below surface of deckboards.

Deviation in dimension of assembled pallets shall be limited to

3/8" out—of-square (3/ü" difference in diagonals), plus or minus 3/16"

in overall pallet length or width.

No combination of defects in workmanship, including nail splits, which will adversely affect the strength of the pallet to a material extent will be permitted.

Source: Eichler, J. R. 1976. Wood Pallet Manufacturing Practices.

Eichler Associates: Cape Coral, Fla.

REGIONAL UTILIZATION OF REUSABLE PALLETS BY THE GROCERY AND RELATED PRODUCTS INDUSTRY by

Robert Bruce Anderson

_ Committee Chairman: Harold W. Wisdom ‘ ' Forestry and Forest Products

(ABSTRACT)

Since 1960, pallet production has quadrupled, increasing the

pallet industry's use of hardwood lumber from lk percent to almost 50

percent of total hardwood lumber production. Part of this growth can

be attributed to the grocery and related products industry, which

should continue as a major growth area for pallet usage over the next decade.

The general objective of this study is to provide information

that can be used to understand the long-term potential and long-term

trends in the grocery pallet market which relate to future regional

timber demands by the pallet industry. Specific objectives are:

(A) Provide information on current use of grocery pallets in the grocery distribution industry; (B) Provide theoretical framework for future analysis of the regional demand for grocery pallets; and (C)

Provide information on demand for regional timber resources resulting from grocery pallet production within specified regions.

Models are presented representing demand and supply in the grocery and related products and grocery pallet markets. In the grocery pallet model, demand for new grocery pallets is expressed as an 'excess demand' where demand for new grocery pallets equals the difference between aggregate supply of pallets to grocery distribution and available inventory of grocery pallets in the system. Inventory of grocery pallets in grocery distribution is expressed as a function of dollar volume of retail sales, based on application of a stock adjustment model for durable inputs.

Consumption of grocery pallets by the grocery distribution industry is shown to be an important part in overall new pallet production even though the pallet used, ü8"xNO", only constitutes about

11 percent of total new pallet production in 1986. Estimates of national consumption of new grocery pallets in 1986 are broken down into regional estimates of new pallet consumption.

Volume of wood raw material used in 1986 for production of grocery pallets is estimated to exceed 838 million board foot of wood raw material, or potentially 18 percent of·tota1 hardwood raw material consumed in production of all types of pallets. National trends effecting wood use in grocery distribution are considered. Specific regional trends effecting wood raw material use are not identified.